Vital sign real-time monitoring method and system based on intelligent wearable underwear

Abstract

The invention provides a vital sign real-time monitoring method and system based on intelligent wearable underwear, and the method comprises the steps: obtaining the vital sign data of a user based on a sensor integrated with the intelligent wearable underwear; acquiring the current motion activity state of the user through an accelerometer and a gyroscope which are arranged in the intelligent wearable underwear, and acquiring a vital sign data threshold value of the user based on the current motion activity state of the user, the vital sign data change of the user in a preset time window and the vital sign data of the user in the same historical time period; when the vital sign data of the user exceed a threshold value, giving out early warning, obtaining abnormal vital sign data of the user giving out the early warning, training an abnormity prediction model based on the abnormal vital sign data of the user, and predicting the abnormal condition of the vital signs of the user based on the abnormity prediction model. Through the method and the corresponding system, the vital sign data threshold value is customized for each user by considering the state of the user, and the monitoring accuracy is improved.

Description

Technical Field

[0001] The present invention proposes a method and system for real-time monitoring of vital signs based on smart wearable underwear, and relates to the technical field of vital signs monitoring. Background Art

[0002] For vital signs monitoring, accurate threshold setting is crucial, however, the current method of using the same threshold has many defects. First, individual differences are seriously ignored. Everyone's physical condition, physiological characteristics and living habits are different, and the same threshold cannot adapt to the actual situation of different individuals. For example, the basal heart rate of athletes is usually lower than that of ordinary people. If the same threshold is used, athletes' heart rate abnormalities may be frequently falsely reported. For some patients with chronic diseases, the normal range of their vital signs may be quite different from that of healthy people. The same threshold is difficult to accurately reflect their true status, resulting in an increased risk of underreporting.

[0003] In addition, the range of changes in vital signs of the human body varies greatly under different activity states. When people are in motion, their heart rate and breathing rate will increase significantly, while they will be relatively stable when they are resting quietly. The same threshold cannot be dynamically adjusted according to the real-time changes in activity status, which means that during exercise, normal increases in vital signs may be mistakenly identified as abnormalities, or when resting, minor abnormalities that have occurred may be ignored because they do not reach the fixed threshold, thus missing the opportunity for early intervention and affecting the timeliness and effectiveness of health monitoring. Summary of the invention

[0004] The present invention provides a method and system for real-time monitoring of vital signs based on smart wearable underwear to solve the above-mentioned problems: The present invention proposes a method for real-time monitoring of vital signs based on smart wearable underwear, the method comprising: Obtaining the user's vital signs data based on sensors integrated into smart wearable underwear; The user's current sports activity status is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the user's vital sign data threshold is obtained based on the user's current sports activity status, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period; When the user's vital sign data exceeds a threshold, an early warning is issued, abnormal vital sign data of the user that issued the early warning is obtained, an abnormal prediction model is trained based on the abnormal vital sign data of the user, and abnormal vital sign conditions of the user are predicted based on the abnormal prediction model.

[0005] Furthermore, the user's vital signs data are obtained through the smart wearable underwear, including: The flexible electrocardiogram sensor integrated in the chest position of the smart wearable underwear collects the user's heart rate data according to a preset collection frequency; The piezoelectric sensor integrated in the abdomen of the smart wearable underwear collects the user's breathing rate data according to a preset collection frequency; The collected heart rate data and respiratory rate data are sent to a central processor integrated in the smart wearable underwear.

[0006] Furthermore, the user's current state is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the user's vital sign data threshold is obtained based on the user's current state, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period, including: The user's current activity state is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the activity state includes: stillness, walking, jogging and sprinting; The conductivity data of the user's skin is obtained through the built-in skin conductivity sensor of the smart wearable underwear; Obtain the respiratory rate and heart rate data of the same historical time period by integrating EEG sensors at the collar and shoulders of the smart wearable underwear to obtain the amplitude of the user's EEG signal; Determine the user's sleeping posture through the built-in accelerometer of the smart wearable underwear; The user's vital sign thresholds are obtained based on the vital sign threshold model.

[0007] Furthermore, obtaining the user's vital sign threshold based on the vital sign threshold model includes: Acquire a user's heart rate threshold and a breathing rate threshold based on a heart rate threshold model and a breathing rate threshold model; Specifically, the heart rate threshold model is: in, represents the user's heart rate threshold at time t, , and represents the weight coefficient, + + =1, n represents the time window length for calculating the current heart rate trend, represents the heart rate at time i, m represents the number of time periods considering the heart rate of the same time period in the past, represents the time period weight coefficient, represents the heart rate in the same period in the past, ΔS represents the change in skin conductivity at time t relative to the previous moment, represents the standard deviation of skin conductivity variation, represents the user's basic heart rate, β represents the activity influence coefficient, represents the weight of each activity type, represents the proportion of each activity type at time t, and s represents the number of activity types; The respiratory rate threshold model is:

[0008] in, represents the user's breathing rate threshold at time t, , and represents the weight coefficient, + + =1, represents the respiratory frequency at time i, p represents the time window length for calculating the current respiratory frequency trend, q represents the number of cycles of respiratory frequency considering the same sleep cycle in the past, represents the period weight coefficient, Indicates the breathing rate of the same sleep cycle in the past. Indicates the amplitude change of the EEG signal at time t relative to the previous time. Indicates the standard deviation of the EEG signal amplitude change, Indicates the user's basic breathing rate, represents the influence coefficient of sleeping posture, represents the weight of each sleeping posture, It represents the proportion of each sleeping posture at time t, and r represents the number of sleeping posture types, including supine, side-lying and prone.

[0009] Furthermore, when the user's vital sign data exceeds a threshold, an early warning is issued, abnormal vital sign data of the user that issued the early warning is obtained, and an abnormal prediction model is trained based on the abnormal vital sign data of the user, including: When the user's vital sign data exceeds a threshold, an early warning is issued, and at the same time, abnormal vital sign data of the user within a time window when the early warning is issued is obtained, and features in the abnormal vital sign data are extracted, wherein the features include: average heart rate, maximum heart rate, minimum heart rate, rising slope of heart rate, falling slope of heart rate, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle, and standard deviation of respiratory rate within a respiratory cycle; The abnormal vital sign data are divided into a training set and a test set, and the long short-term memory network model is trained using the data in the training set, and the parameters are adjusted to the optimal value; Receive new vital sign data in real time, input the vital sign data into the pre-trained model, and the long short-term memory network model predicts whether the vital sign data has abnormal conditions based on the input real-time data, and issues an early warning if an abnormality is predicted.

[0010] The present invention proposes a real-time vital sign monitoring system based on smart wearable underwear, the system comprising: A module for acquiring vital signs data, which acquires the user's vital signs data based on sensors integrated in the smart wearable underwear; A module for obtaining a vital sign data threshold value is used to obtain the user's current motion activity state through the accelerometer and gyroscope built into the smart wearable underwear, and obtain the user's vital sign data threshold value based on the user's current motion activity state, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period; The abnormality prediction module is used to issue an early warning when the user's vital signs data exceeds a threshold, obtain the abnormal vital signs data of the user for issuing the early warning, train an abnormality prediction model based on the abnormal vital signs data of the user, and predict the abnormal situation of the user's vital signs based on the abnormality prediction model.

[0011] Furthermore, the module for acquiring vital signs data includes: A heart rate data collection module, which is used for collecting the user's heart rate data according to a preset collection frequency through a flexible electrocardiogram sensor integrated in the chest position of the smart wearable underwear; A respiratory frequency data collection module, which is used for collecting the user's respiratory frequency data according to a preset collection frequency through a piezoelectric sensor integrated in the abdomen of the smart wearable underwear; The sending module is used to send the collected heart rate data and respiratory rate data to a central processor integrated in the smart wearable underwear.

[0012] Furthermore, the module for obtaining the threshold value of vital sign data includes: The activity status acquisition module is used to acquire the user's current activity status through the accelerometer and gyroscope built into the smart wearable underwear, and the activity status includes: stillness, walking, jogging and sprinting; A conductivity data acquisition module is used to acquire the conductivity data of the user's skin through the built-in skin electrical sensor of the smart wearable underwear; The module for obtaining the amplitude of the EEG signal is used to obtain the respiratory rate and heart rate data of the same historical time period by integrating EEG sensors at the collar and shoulders of the smart wearable underwear to obtain the amplitude of the user's EEG signal; The sleeping posture determination module is used to determine the user's sleeping posture through the accelerometer built into the smart wearable underwear; The vital sign threshold calculation module is used to obtain the user's vital sign threshold based on the vital sign threshold model.

[0013] Furthermore, the module for calculating the vital sign threshold value includes: A heart rate threshold calculation module is used to obtain a user's heart rate threshold and a respiratory rate threshold based on a heart rate threshold model and a respiratory rate threshold model; Specifically, the heart rate threshold model is: in, represents the user's heart rate threshold at time t, , and represents the weight coefficient, + + =1, n represents the time window length for calculating the current heart rate trend, represents the heart rate at time i, m represents the number of time periods considering the heart rate of the same time period in the past, represents the time period weight coefficient, represents the heart rate in the same period in the past, ΔS represents the change in skin conductivity at time t relative to the previous moment, represents the standard deviation of skin conductivity variation, represents the user's basic heart rate, β represents the activity influence coefficient, represents the weight of each activity type, represents the proportion of each activity type at time t, and s represents the number of activity types; Calculate the respiratory rate threshold module. Specifically, the respiratory rate threshold model is:

[0014] in, represents the user's breathing rate threshold at time t, , and represents the weight coefficient, + + =1, represents the respiratory frequency at time i, p represents the time window length for calculating the current respiratory frequency trend, q represents the number of cycles of respiratory frequency considering the same sleep cycle in the past, represents the period weight coefficient, Indicates the breathing rate of the same sleep cycle in the past. Indicates the amplitude change of the EEG signal at time t relative to the previous time. Indicates the standard deviation of the EEG signal amplitude change, Indicates the user's basic breathing rate, represents the influence coefficient of sleeping posture, represents the weight of each sleeping posture, It represents the proportion of each sleeping posture at time t, and r represents the number of sleeping posture types, including supine, side-lying and prone.

[0015] Furthermore, the abnormality prediction module includes: A feature extraction module is used to issue an early warning when the user's vital sign data exceeds a threshold value, and simultaneously obtain abnormal vital sign data of the user within a time window when the early warning is issued, and extract features from the abnormal vital sign data, wherein the features include: average heart rate, maximum heart rate, minimum heart rate, rising slope of heart rate, falling slope of heart rate, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle, and standard deviation of respiratory rate within a respiratory cycle; A training module, used to divide the abnormal vital sign data into a training set and a test set, use the data in the training set to train the long short-term memory network model, and adjust the parameters to the optimal level; The prediction module is used to receive new vital sign data in real time and input the vital sign data into the pre-trained model. The long short-term memory network model predicts whether there are any abnormalities in the vital sign data based on the input real-time data, and issues an early warning if an abnormality is predicted.

[0016] Beneficial effects of the present invention: accurate personalized monitoring, traditional vital signs monitoring often uses universal standard thresholds, ignoring individual differences. This technical solution fully considers the user's own movement status, recent vital signs fluctuations and historical data characteristics, and tailors the vital signs data thresholds for each user. This is like creating an exclusive "health ruler" for each person, which greatly improves the accuracy of monitoring and can more keenly capture subtle abnormalities in each user's vital signs to avoid misjudgment and missed judgments; timely risk warning, real-time dynamic threshold comparison and instant warning mechanism, to ensure that an alarm is issued as soon as an abnormal vital sign occurs, which is like setting a 24-hour uninterrupted "warning line" for the user's health, which can promptly detect potential health risks, win precious treatment time for users or remind users to adjust their lifestyles in time to prevent further development of the disease; predictive health management, through deep learning of historical abnormal data, the abnormal prediction model has the ability to predict future vital signs abnormalities. This function has achieved a major transformation in health management from "after-the-fact remedy" to "pre-emptive prevention", helping users and medical personnel to develop personalized health management plans in advance, take targeted preventive measures, reduce the probability of disease occurrence, and improve overall health levels; convenient and continuous monitoring, smart wearable underwear as a carrier for data collection, is highly convenient and comfortable, and users can continue to wear it in various scenarios such as daily life, work, and exercise, without the need for additional complex operations or specific environmental requirements. This uninterrupted data collection method ensures the consistency and integrity of vital sign data, provides a rich data foundation for a comprehensive and in-depth understanding of the user's health status, and helps to discover some hidden health problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of a method for real-time monitoring of vital signs based on smart wearable underwear according to the present invention. DETAILED DESCRIPTION

[0018] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0019] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0021] One embodiment of the present invention provides a method for real-time monitoring of vital signs based on smart wearable underwear, the method comprising: Obtaining the user's vital signs data based on sensors integrated into smart wearable underwear; The user's current sports activity status is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the user's vital sign data threshold is obtained based on the user's current sports activity status, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period; When the user's vital sign data exceeds a threshold, an early warning is issued, abnormal vital sign data of the user that issued the early warning is obtained, an abnormal prediction model is trained based on the abnormal vital sign data of the user, and abnormal vital sign conditions of the user are predicted based on the abnormal prediction model.

[0022] The working principle and effect of the above technical solution are as follows: vital sign data collection, the smart wearable underwear integrates a variety of sensors, these sensors are like sharp "scouts", which can capture various vital sign data of users in real time, such as heart rate, respiratory rate, body temperature, etc.; they convert human physiological signals into electrical signals or digital signals, and provide raw data support for subsequent analysis; motion state data collection, accelerometers and gyroscopes are built into smart wearable underwear, by sensing the acceleration changes and angular rotations generated when the human body moves, to accurately judge the user's current motion activity state, the accelerometer can judge whether the user is stationary, walking or running according to the acceleration amplitude and frequency during exercise, and the gyroscope can assist in determining the movement posture and movement changes by detecting the angular velocity of the body rotation; different motion activity states have a significant impact on vital signs, for example, when the accelerometer and gyroscope judge that the user is in a running state, according to physiological common sense, the heart rate and respiratory rate are usually higher than the stationary state, so the current motion activity state is an important reference factor for determining the threshold of vital signs; recent data changes: within the preset time window, analyze the change trend and amplitude of vital sign data, for example, observe the past 5 Whether the heart rate gradually rises, falls, or remains stable within a minute, as well as the rate of rise or fall, such recent data changes reflect the dynamic situation of vital signs and help to set thresholds more accurately; the vital signs data of the same historical time period of the user contains the individual's physiological laws and normality. For example, long-term monitoring has found that the user's heart rate between 10 am and 11 am every day usually fluctuates within a certain range; when determining the vital signs threshold for the current time period, referring to historical data can make the threshold more in line with the user's individual characteristics; by combining these three factors and using specific algorithms, the vital signs data threshold that adapts to the user's current condition is calculated; the vital signs data collected in real time is compared with the calculated threshold in real time. Once the vital signs data crosses the threshold red line, the system immediately activates the early warning mechanism, just like sounding an alarm, indicating that an abnormal situation may have occurred; at the moment of issuing the early warning, the system quickly records the abnormal vital signs data of the user at this time, and these data become the key materials for subsequent in-depth analysis and model training; the collected abnormal vital signs data of the user is used as a "teaching material" to train the abnormal prediction model. During the learning process, the model mines the hidden patterns, features and laws in the data, and gradually establishes the internal connection between vital signs data and abnormal situations; the fully trained abnormal prediction model can analyze and judge the newly input vital signs data, predict whether the user's vital signs will be abnormal in the future, and provide a basis for early intervention and prevention; accurate and personalized monitoring, traditional vital signs monitoring often uses a universal standard threshold, ignoring individual differences. This technical solution fully considers the user's own movement status, recent vital signs fluctuations and historical data characteristics, and tailors the vital signs data threshold for each user.This is like creating a unique "health ruler" for everyone, which greatly improves the accuracy of monitoring and can more keenly capture the subtle abnormalities of each user's vital signs to avoid misjudgment and missed judgment; timely risk warning, real-time dynamic threshold comparison and instant warning mechanism, ensure that the alarm is issued as soon as the vital signs are abnormal, which is like setting a 24-hour uninterrupted "warning line" for the user's health, which can timely discover potential health risks, win precious treatment time for users or remind users to adjust their lifestyle in time to prevent the further development of diseases; predictive health management, through deep learning of historical abnormal data, the abnormal prediction model has the ability to predict future abnormal vital signs. This function realizes a major transformation of health management from "after-the-fact remedy" to "pre-prevention", helping users and medical personnel to formulate personalized health management plans in advance, take targeted preventive measures, reduce the probability of disease occurrence, and improve overall health level; convenient and continuous monitoring, smart wearable underwear as a carrier of data collection is highly convenient and comfortable, and users can wear it continuously in various scenarios such as daily life, work, and sports, without additional complex operations or specific environmental requirements. This uninterrupted data collection method ensures the consistency and integrity of vital signs data, provides a rich data foundation for a comprehensive and in-depth understanding of the user's health status, and helps to discover some hidden health problems.

[0023] In one embodiment of the present invention, obtaining vital sign data of a user through smart wearable underwear includes: The flexible electrocardiogram sensor integrated in the chest position of the smart wearable underwear collects the user's heart rate data according to a preset collection frequency; The piezoelectric sensor integrated in the abdomen of the smart wearable underwear collects the user's breathing rate data according to a preset collection frequency; The collected heart rate data and respiratory rate data are sent to a central processor integrated in the smart wearable underwear.

[0024] The working principle and effect of the above technical solution are as follows: heart rate data collection, the flexible ECG sensor is cleverly integrated in the chest of the smart wearable underwear. The chest is the area where heart activity is most obvious, and the weak signal generated by the heart's electrical activity can be effectively captured here. The sensor periodically detects the heart's electrical activity according to the preset acquisition frequency, and converts the electrical signal generated by the heartbeat into heart rate data that can be recorded and processed. For example, if the preset acquisition frequency is 1000 times per second, the sensor will sample the heart's electrical activity 1000 times per second, thereby accurately obtaining real-time information on the heartbeat; respiratory rate data collection, the piezoelectric sensor is installed in the abdomen of the smart wearable underwear. When the human body breathes, the abdomen will produce regular ups and downs with the breathing movement. The piezoelectric sensor can sense this tiny pressure change. When the abdomen squeezes or stretches the sensor due to breathing, the sensor converts the pressure change into an electrical signal based on the piezoelectric effect, and then converts it into respiratory rate data according to the preset acquisition frequency. Similarly, if the preset acquisition frequency is 50 times per second, the sensor will sample the abdominal pressure change 50 times per second. The detection is performed to obtain the real-time situation of the respiratory rate; the collected heart rate data and respiratory rate data are sent to the central processor integrated in the smart wearable underwear through the preset circuit or wireless transmission module inside the smart wearable underwear, providing a basis for subsequent analysis and judgment. Accurate real-time monitoring, the sensor setting at a specific position on the smart wearable underwear makes the collection of heart rate and respiratory rate data more accurate. The flexible ECG sensor at the chest position can directly and accurately capture the electrical activity of the heart and obtain high-precision heart rate data; the piezoelectric sensor at the abdomen is extremely sensitive to the changes in abdominal pressure caused by breathing, and can monitor the respiratory rate in real time and accurately. At the same time, the preset collection frequency ensures the high timeliness of the data and can reflect the dynamic changes of heart rate and respiratory rate in a timely manner; integrating the sensor into the smart wearable underwear makes the monitoring process comfortable and convenient. Patients do not need to wear additional equipment that may affect their daily life. They only need to wear underwear normally to complete data collection. This method improves the patient's acceptance and compliance, ensures the continuity and stability of data collection, and helps to monitor the patient's physiological state in a long-term and comprehensive manner; for epileptic patients, heart rate and respiratory rate may change abnormally before an epileptic attack. Continuous and accurate monitoring of these two vital signs provides important data support for the early prediction of epilepsy. Based on these real-time data, the central processor can combine specific algorithms and models to analyze the changing patterns of heart rate and respiratory rate, and try to find the rules related to epileptic seizures, thereby achieving early warning of epileptic seizures, buying patients precious response time, and reducing the risks brought by epileptic seizures.

[0025] In one embodiment of the present invention, the current state of the user is obtained through the accelerometer and gyroscope built into the smart wearable underwear, and the vital sign data threshold of the user is obtained based on the current state of the user, the change of the vital sign data of the user within a preset time window, and the vital sign data of the user in the same historical time period, including: The user's current activity state is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the activity state includes: stillness, walking, jogging and sprinting; The conductivity data of the user's skin is obtained through the built-in skin conductivity sensor of the smart wearable underwear; Obtain the respiratory rate and heart rate data of the same historical time period by integrating EEG sensors at the collar and shoulders of the smart wearable underwear to obtain the amplitude of the user's EEG signal; Determine the user's sleeping posture through the built-in accelerometer of the smart wearable underwear; The user's vital sign thresholds are obtained based on the vital sign threshold model.

[0026] The working principle and effect of the above technical solution are as follows: activity status monitoring, the accelerometer and gyroscope built into the smart wearable underwear work together, the accelerometer senses the strength and rhythm of body movement by detecting acceleration changes in different directions; the gyroscope is responsible for monitoring the body's rotation angle and angular velocity. When the acceleration detected by the accelerometer fluctuates within a very small range and the angular velocity measured by the gyroscope is close to zero, the user is judged to be in a stationary state; when periodic and specific frequency and amplitude acceleration changes are detected, combined with the body swing sensed by the gyroscope, it can be identified as a walking state; if the frequency and amplitude of the acceleration changes increase, and the body rotation is more obvious, it is determined to be jogging; when the acceleration and angular velocity changes are more drastic and rapid, it is determined to be a sprint state; skin conductivity data collection, the skin conductivity sensor is close to the user's skin, and the conductivity of the skin surface is measured using the principle of skin electrical activity. When the user's emotions or physiological state changes, the sympathetic nerves are excited, The sweat glands are stimulated to secrete, and the electrolytes in the sweat cause the skin conductivity to change. The sensor captures these changes in real time. From the data storage module of the smart wearable underwear or the cloud database associated with it, the respiratory rate and heart rate data of the same historical time period are extracted according to the current time information to provide a historical reference for analyzing the current vital signs. The EEG signal amplitude collection, the EEG sensor integrated in the collar and shoulder of the smart wearable underwear contacts the skin through electrodes, collects the weak electrical signals generated by the brain's neuronal activity and transmitted to the surface of the scalp. After amplification, filtering and other processing, these signals are processed to obtain the amplitude data of the EEG signal, which reflects the brain's sleep stage. Sleep posture judgment, the accelerometer continuously monitors the acceleration of the body in different directions while the user is sleeping. For example, when lying on your back, the acceleration in all directions of the body is relatively stable and conforms to a specific pattern; when lying on your side, the acceleration on one side will change significantly; when lying prone, the acceleration distribution presents different characteristics. The user's sleeping posture is determined by analyzing the patterns of these acceleration data. The vital signs threshold model comprehensively considers multiple factors such as the user's current activity status, skin conductivity data, respiratory rate and heart rate data in the same historical time period, EEG signal amplitude, and sleeping posture. Through preset algorithms and weight distribution, these data are analyzed and calculated to obtain the vital signs thresholds, heart rate thresholds, and respiratory rate thresholds suitable for the user's current condition.Comprehensive and accurate monitoring: Through a variety of sensors, data is collected from multiple dimensions such as activity status, galvanic skin response, EEG activity, sleeping posture, etc., and combined with historical data, the user's physiological and behavioral status is monitored comprehensively and accurately, providing a rich and detailed data basis for accurately assessing the user's health status; personalized threshold setting: The vital sign threshold model determines the threshold based on multi-source data, fully considering individual differences. Compared with the general standard, it can better fit the actual situation of each user, making vital sign monitoring more targeted and accurate, and improving the reliability of abnormal situation judgment; early risk warning: accurate monitoring and personalized threshold setting can help to timely detect abnormal changes in the user's vital signs and warn of potential health risks in advance. For example, for epilepsy patients, combined with EEG signal amplitude and other physiological data, physical changes before the onset may be detected earlier, which can help patients and their caregivers to cope with various sensors integrated into smart wearable underwear. Users do not need to wear multiple devices additionally, and can complete various data collection during daily wear, which improves the convenience and comfort of monitoring, increases user willingness and compliance, and ensures the continuity and stability of data collection.

[0027] In one embodiment of the present invention, obtaining a user's vital sign threshold based on a vital sign threshold model includes: Acquire a user's heart rate threshold and a breathing rate threshold based on a heart rate threshold model and a breathing rate threshold model; Specifically, the heart rate threshold model is: in, represents the user's heart rate threshold at time t, , and represents the weight coefficient, + + =1, n represents the time window length for calculating the current heart rate trend, represents the heart rate at time i, m represents the number of time periods considering the heart rate of the same time period in the past, represents the time period weight coefficient, represents the heart rate in the same period in the past, ΔS represents the change in skin conductivity at time t relative to the previous moment, represents the standard deviation of skin conductivity variation, represents the user's basic heart rate, β represents the activity influence coefficient, represents the weight of each activity type, represents the proportion of each activity type at time t, and s represents the number of activity types; The respiratory rate threshold model is:

[0028] in, represents the user's breathing rate threshold at time t, , and represents the weight coefficient, + + =1, represents the respiratory frequency at time i, p represents the time window length for calculating the current respiratory frequency trend, q represents the number of cycles of respiratory frequency considering the same sleep cycle in the past, represents the period weight coefficient, Indicates the breathing rate of the same sleep cycle in the past. Indicates the amplitude change of the EEG signal at time t relative to the previous time. Indicates the standard deviation of the EEG signal amplitude change, Indicates the user's basic breathing rate, represents the influence coefficient of sleeping posture, represents the weight of each sleeping posture, It represents the proportion of each sleeping posture at time t, and r represents the number of sleeping posture types, including supine, side-lying and prone.

[0029] The working principle and effect of the above technical solution are as follows: the heart rate threshold model comprehensively considers multiple factors that affect the heart rate to more accurately determine the user's heart rate threshold at different times. These factors include the current heart rate trend, heart rate in the same period in the past, skin conductivity changes, basal heart rate and activity status. Skin conductivity changes reflect user stress, and stress will also affect the user's heart rate. By incorporating these factors into the formula, it can more comprehensively reflect the impact of the user's physiological state and behavior on the heart rate, avoiding errors and inaccuracies caused by a single factor. Determining the threshold based only on the current heart rate may not take into account the individual physiological differences of the user and the long-term law of heart rate changes; and only referring to past heart rate data may ignore special circumstances at the current moment, such as sudden emotional changes or physical activities. Therefore, the comprehensive multi-factor design is to more accurately adapt to different scenarios and individual differences. Multiple weight coefficients are introduced in the formula to reflect the differences in the importance of different factors in determining the heart rate threshold. This design is based on an in-depth understanding and data analysis of the relationship between each factor and heart rate. For example, the basal heart rate may have a relatively stable effect on the overall heart rate threshold, so it is given a certain weight; and the current heart rate trend (through The purpose of setting the time window length is to capture the changing trend of heart rate within a certain time range, without paying too much attention to short-term accidental fluctuations, and without ignoring important recent changes. For example, a shorter time window may be affected by momentary interference factors, while a longer time window may not be able to reflect rapid changes in heart rate in a timely manner; the purpose of considering the number of time periods for heart rate in the same time period in the past is to mine the long-term regularity and periodic characteristics of the user's heart rate from historical data, by taking a weighted average of multiple heart rate data from the same time period in the past (through embodied), can better understand the user's heart rate normality in a specific time period, and provide a reference for determining the current heart rate threshold. This design helps the model adapt to the living habits and physiological rhythms of different users. Personalized and accurate monitoring, because it comprehensively considers factors such as the user's basic heart rate, current heart rate trend, heart rate in the same period in the past, skin conductivity changes and activity status, and fine-tunes parameters such as weight allocation and time window, this formula can provide each user with a personalized heart rate threshold. Different users have different physiological characteristics, lifestyles and activity habits. This personalized monitoring method can more accurately capture the normal range and abnormal changes of each user's heart rate, avoiding misjudgments and omissions that may be caused by general thresholds; for example, for a user who exercises frequently, his basic heart rate may be low, and his heart rate rises significantly during exercise. The model will give a more realistic prediction based on his historical exercise data and current activity status. The heart rate threshold for different situations is set according to the actual situation, rather than simply adopting the standard for the general population; the formula includes parameters such as the real-time skin conductivity change value and the proportion of each activity type at the current moment, so that the heart rate threshold can be adjusted dynamically in real time according to the user's current physiological and behavioral state. When the user's skin conductivity changes suddenly (which may reflect emotional fluctuations or physical stress response) or the activity state changes (such as from stillness to movement), the heart rate threshold will be updated quickly accordingly to adapt to these changes; the ability to adjust dynamically in real time is very important for timely detection of heart rate abnormalities. For example, when a user suddenly engages in strenuous exercise, the heart rate will rise rapidly. If the heart rate threshold cannot be adjusted in time, a normal heart rate increase may be mistakenly judged as abnormal. The formula can adjust the threshold in real time according to the change of activity status, ensuring that the heart rate can be accurately monitored in various situations; improve the reliability and effectiveness of health monitoring. By integrating multiple factors, personalization and real-time dynamic adjustment, the heart rate threshold model can significantly improve the reliability and effectiveness of health monitoring. It can more accurately identify abnormal heart rates, provide users with timely health warnings, help users and medical professionals better understand users' cardiovascular health status, and develop more reasonable health management and treatment plans; for example, for patients with cardiovascular disease, accurate heart rate monitoring can help doctors evaluate the effectiveness of drug treatment, adjust treatment plans, and prevent possible cardiovascular events. For ordinary users, this model can be used for daily health monitoring, timely identify potential health problems, and promote the formation of a healthy lifestyle; the design of the heart rate threshold formula comprehensively considers a variety of factors, and through reasonable weight allocation, time window setting, and real-time parameter inclusion, it realizes personalized, precise, and real-time dynamic heart rate monitoring, which has important practical application value and health monitoring significance.The respiratory rate threshold formula is designed to comprehensively consider the various factors that affect the respiratory rate, so as to more accurately determine the user's respiratory rate threshold at different times. These factors include the current respiratory rate trend, the respiratory rate of the same sleep cycle in the past, the change in the amplitude of the EEG signal, the basic respiratory rate, and the sleeping posture. Combining these factors can more comprehensively reflect the impact of the user's physiological state and behavior on the respiratory rate, and avoid the inaccuracy and limitations caused by a single factor. For example, determining the threshold based only on the current respiratory rate cannot take into account the long-term breathing patterns and physiological differences of individual users; and only referring to the respiratory rate data of past sleep cycles may ignore the special circumstances at the current moment, such as the impact of sudden emotional changes or physical activities on breathing. Therefore, the design of comprehensive multi-factors is to adapt to different scenarios and individual differences more accurately; multiple weight coefficients are introduced in the formula to reflect the importance of different factors in determining the respiratory rate threshold. This design is based on an in-depth understanding and data analysis of the relationship between each factor and the respiratory rate. For example, the basic respiratory rate may have a relatively stable effect on the overall respiratory rate threshold, so it is given a certain weight; and the current respiratory rate trend (through. The time window length is set to capture the changing trend of the respiratory frequency within a certain time range, without paying too much attention to short-term accidental fluctuations, nor ignoring important recent changes. For example, a shorter time window may be affected by momentary interference factors, while a longer time window may not be able to reflect rapid changes in the respiratory frequency in a timely manner. Considering the number of cycles of the respiratory frequency in the same sleep cycle in the past is to mine the long-term regularity and periodic characteristics of the user's respiratory frequency from historical data, especially in the sleep state. By taking a weighted average of the respiratory frequency data of multiple past sleep cycles (through embodied), can better understand the user's breathing normality during sleep, and provide a reference for determining the current breathing rate threshold. This design helps the model adapt to the personalized and accurate monitoring of the sleep habits and physiological rhythms of different users; due to the comprehensive consideration of the user's basic breathing rate, current breathing rate trend, breathing rate in the same sleep cycle in the past, changes in EEG signal amplitude and sleeping posture, and fine-tuning through parameters such as weight distribution and time window, the formula can provide a personalized breathing rate threshold for each user. Different users have different physiological characteristics, lifestyles and sleeping habits. This personalized monitoring method can more accurately capture the normal range and abnormal changes in each user's breathing rate, avoiding misjudgment and missed judgments that may be caused by universal thresholds. For example, for a user with sleep apnea syndrome, his breathing rate during sleep may have an abnormal pattern. The model will give a breathing rate threshold that is more in line with his actual situation based on his historical sleep data and current EEG signals, rather than simply adopting the standards of the general population; the formula includes parameters such as the real-time EEG signal amplitude change value and the proportion of each sleeping posture at the current moment, so that the breathing rate threshold can be adjusted dynamically in real time according to the user's current physiological and behavioral state. When the user's EEG signal amplitude changes suddenly (which may reflect a change in brain activity, such as from light sleep to deep sleep) or the sleeping posture changes (such as from supine to side lying), the respiratory rate threshold will be updated quickly to adapt to these changes. This ability to adjust dynamically in real time is very important for timely detection of abnormal respiratory rate. For example, when the user's sleeping posture changes and causes airway compression, which may affect the respiratory rate, the model can adjust the threshold in real time according to the change in sleeping posture to ensure accurate monitoring of respiratory rate in various situations and timely detection of potential respiratory problems; the EEG amplitude in the deep sleep stage is relatively high; while the amplitude in the shallow sleep stage is relatively low and the fluctuation is small. During the transition of the sleep cycle, the change in amplitude is also an important basis for judging the sleep stage. For example, when going from light sleep to deep sleep, the amplitude usually has a significant increase trend; through the comprehensive multi-factor, personalization and real-time dynamic adjustment, the respiratory rate threshold model can significantly improve the reliability and effectiveness of health monitoring. It can more accurately identify abnormal respiratory rate, provide users with timely health warnings, help users and medical professionals better understand users' respiratory health status, and develop more reasonable health management and treatment plans; for patients with respiratory diseases, accurate respiratory rate monitoring can help doctors evaluate the progression of the disease, adjust treatment plans, and prevent possible serious complications such as respiratory failure.For ordinary users, this model can be used for daily health monitoring, timely detection of potential respiratory problems, and promotion of a healthy lifestyle, such as improving sleeping posture. The design of the respiratory rate threshold formula takes into account a variety of factors, and through reasonable weight allocation, time window setting and real-time parameter incorporation, it achieves personalized, precise and real-time dynamic respiratory rate monitoring, which has important practical application value and health monitoring significance.

[0030] In one embodiment of the present invention, when the user's vital sign data exceeds a threshold, an early warning is issued, abnormal vital sign data of the user issuing the early warning is obtained, and an abnormal prediction model is trained based on the abnormal vital sign data of the user, including: When the user's vital sign data exceeds a threshold, an early warning is issued, and at the same time, abnormal vital sign data of the user within a time window when the early warning is issued is obtained, and features in the abnormal vital sign data are extracted, wherein the features include: average heart rate, maximum heart rate, minimum heart rate, rising slope of heart rate, falling slope of heart rate, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle, and standard deviation of respiratory rate within a respiratory cycle; The abnormal vital sign data are divided into a training set and a test set, and the long short-term memory network model is trained using the data in the training set, and the parameters are adjusted to the optimal value; Receive new vital sign data in real time, input the vital sign data into the pre-trained model, and the long short-term memory network model predicts whether the vital sign data has abnormal conditions based on the input real-time data, and issues an early warning if an abnormality is predicted.

[0031] The working principle and effect of the above technical solution are as follows: data monitoring and early warning triggering, continuously obtaining the user's vital signs data, such as heart rate, respiratory rate, etc., through smart wearable devices or other monitoring means. When these vital signs data exceed the preset threshold, the system immediately issues a warning signal and records the abnormal user vital signs data within the time window; extracts a series of features from the acquired abnormal vital signs data, including average heart rate, maximum heart rate, minimum heart rate, heart rate rising slope, heart rate falling slope, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle and standard deviation of respiratory rate within the respiratory cycle, etc. These features can comprehensively reflect the changes and trends of vital signs and provide rich information for subsequent model training and prediction; the extracted abnormal vital signs data are divided into training set and test set. The training set is used to train the long short-term memory network (LSTM) model. The LSTM It is a special recurrent neural network that is good at processing time series data and can capture long-term dependencies and temporal dynamic changes in the data. During the training process, by continuously adjusting the parameters of the model, such as weights and biases, the model can achieve the best performance on the training set, that is, it can accurately learn the characteristics and patterns of abnormal vital signs data, thereby establishing a mapping relationship between vital signs data and abnormal conditions. It receives new vital signs data in real time and inputs it into the pre-trained long short-term memory network model. The model predicts whether there are abnormalities in the vital signs data based on the input real-time data and the previously learned knowledge and patterns. If an abnormality is predicted, the system will issue an early warning again so that appropriate measures can be taken in time. Accurate prediction and timely early warning, using the long short-term memory network model to analyze and predict vital signs data, can fully consider the time series characteristics and long-term dependencies of vital signs data, so as to more accurately capture the trend and pattern of abnormal conditions. Compared with the traditional warning method based on fixed thresholds, this prediction method based on machine learning models can detect potential abnormalities in advance and buy more time for medical intervention and health management; the real-time prediction and warning functions allow users or medical staff to know the abnormal changes in vital signs at the first time and take timely measures, such as adjusting treatment plans, conducting further examinations or providing emergency assistance, etc., thereby improving the timeliness and effectiveness of medical treatment and reducing health risks; personalized health management, each user's vital signs data has unique characteristics and change patterns. Through feature extraction and model training of individual vital signs abnormal data, a personalized health model can be established for each user.The model can provide health assessment and early warning that is more in line with individual actual conditions based on the user's historical data and real-time monitoring data, and realize personalized health management; for example, for patients with chronic diseases, personalized health models can provide more accurate disease management suggestions based on the patient's disease characteristics and vital signs change trends, helping patients to better control their disease and improve their quality of life; data-driven health insights, in the process of model training and prediction, the analysis and processing of a large amount of vital signs data can dig out some potential health information and laws. This information not only helps to detect abnormal conditions in a timely manner, but also provides data support for doctors and researchers, helping them to deeply understand the occurrence and development mechanism of the disease, the impact of different factors on health, etc., thereby promoting the development of medical research and clinical practice; through the analysis of a large amount of user vital signs data, it may be found that certain specific vital signs change patterns are associated with the early symptoms of a certain disease, which provides new ideas and methods for the early diagnosis and prevention of the disease; improve the efficiency and quality of health monitoring, and the automated data monitoring, feature extraction, model training and prediction and early warning processes greatly reduce the workload of manual intervention and analysis, and improve the efficiency of health monitoring. At the same time, accurate predictions and early warnings based on machine learning models have also improved the quality of health monitoring, reduced false alarms and missed alarms, and made health monitoring more reliable and credible. Efficient and high-quality health monitoring methods can be widely used in hospitals, home care, telemedicine and other scenarios, providing users with all-round, all-weather health protection.

[0032] One embodiment of the present invention provides a real-time vital sign monitoring system based on smart wearable underwear, the system comprising: A module for acquiring vital signs data, which acquires the user's vital signs data based on sensors integrated in the smart wearable underwear; A module for obtaining a vital sign data threshold value is used to obtain the user's current motion activity state through the accelerometer and gyroscope built into the smart wearable underwear, and obtain the user's vital sign data threshold value based on the user's current motion activity state, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period; The abnormality prediction module is used to issue an early warning when the user's vital signs data exceeds a threshold, obtain the abnormal vital signs data of the user for issuing the early warning, train an abnormality prediction model based on the abnormal vital signs data of the user, and predict the abnormal situation of the user's vital signs based on the abnormality prediction model.

[0033] In one embodiment of the present invention, the module for acquiring vital signs data includes: A heart rate data collection module, which is used for collecting the user's heart rate data according to a preset collection frequency through a flexible electrocardiogram sensor integrated in the chest position of the smart wearable underwear; A respiratory frequency data collection module, which is used for collecting the user's respiratory frequency data according to a preset collection frequency through a piezoelectric sensor integrated in the abdomen of the smart wearable underwear; The sending module is used to send the collected heart rate data and respiratory rate data to a central processor integrated in the smart wearable underwear.

[0034] In one embodiment of the present invention, the module for obtaining the threshold value of vital sign data includes: The activity status acquisition module is used to acquire the user's current activity status through the accelerometer and gyroscope built into the smart wearable underwear, and the activity status includes: stillness, walking, jogging and sprinting; A conductivity data acquisition module is used to acquire the conductivity data of the user's skin through the built-in skin electrical sensor of the smart wearable underwear; The module for obtaining the amplitude of the EEG signal is used to obtain the respiratory rate and heart rate data of the same historical time period by integrating EEG sensors at the collar and shoulders of the smart wearable underwear to obtain the amplitude of the user's EEG signal; The sleeping posture determination module is used to determine the user's sleeping posture through the accelerometer built into the smart wearable underwear; The vital sign threshold calculation module is used to obtain the user's vital sign threshold based on the vital sign threshold model.

[0035] In one embodiment of the present invention, the module for calculating the vital sign threshold value includes: A heart rate threshold calculation module is used to obtain a user's heart rate threshold and a respiratory rate threshold based on a heart rate threshold model and a respiratory rate threshold model; Specifically, the heart rate threshold model is: in, represents the user's heart rate threshold at time t, , and represents the weight coefficient, + + =1, n represents the time window length for calculating the current heart rate trend, represents the heart rate at time i, m represents the number of time periods considering the heart rate of the same time period in the past, represents the time period weight coefficient, represents the heart rate in the same period in the past, ΔS represents the change in skin conductivity at time t relative to the previous moment, represents the standard deviation of skin conductivity variation, represents the user's basic heart rate, β represents the activity influence coefficient, represents the weight of each activity type, represents the proportion of each activity type at time t, and s represents the number of activity types; Calculate the respiratory rate threshold module. Specifically, the respiratory rate threshold model is:

[0036] in, represents the user's breathing rate threshold at time t, , and represents the weight coefficient, + + =1, represents the respiratory frequency at time i, p represents the time window length for calculating the current respiratory frequency trend, q represents the number of cycles of respiratory frequency considering the same sleep cycle in the past, represents the period weight coefficient, Indicates the breathing rate of the same sleep cycle in the past. Indicates the amplitude change of the EEG signal at time t relative to the previous time. Indicates the standard deviation of the EEG signal amplitude change, Indicates the user's basic breathing rate, represents the influence coefficient of sleeping posture, represents the weight of each sleeping posture, It represents the proportion of each sleeping posture at time t, and r represents the number of sleeping posture types, including supine, side-lying and prone.

[0037] In one embodiment of the present invention, the abnormality prediction module includes: A feature extraction module is used to issue an early warning when the user's vital sign data exceeds a threshold value, and simultaneously obtain abnormal vital sign data of the user within a time window when the early warning is issued, and extract features from the abnormal vital sign data, wherein the features include: average heart rate, maximum heart rate, minimum heart rate, rising slope of heart rate, falling slope of heart rate, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle, and standard deviation of respiratory rate within a respiratory cycle; A training module, used to divide the abnormal vital sign data into a training set and a test set, use the data in the training set to train the long short-term memory network model, and adjust the parameters to the optimal level; The prediction module is used to receive new vital sign data in real time and input the vital sign data into the pre-trained model. The long short-term memory network model predicts whether there are any abnormalities in the vital sign data based on the input real-time data, and issues an early warning if an abnormality is predicted.

[0038] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A real-time monitoring method of vital signs based on smart wearable underwear, characterized in that: The method comprises: Obtaining the user's vital signs data based on sensors integrated into smart wearable underwear; The user's current sports activity status is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the user's vital sign data threshold is obtained based on the user's current sports activity status, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period; When the user's vital sign data exceeds a threshold, an early warning is issued, abnormal vital sign data of the user that issued the early warning is obtained, an abnormal prediction model is trained based on the abnormal vital sign data of the user, and abnormal vital sign conditions of the user are predicted based on the abnormal prediction model.

2. According to claim 1, a real-time monitoring method of vital signs based on smart wearable underwear is characterized in that: Obtain the user's vital signs data through smart wearable underwear, including: The flexible electrocardiogram sensor integrated in the chest position of the smart wearable underwear collects the user's heart rate data according to a preset collection frequency; The piezoelectric sensor integrated in the abdomen of the smart wearable underwear collects the user's breathing rate data according to a preset collection frequency; The collected heart rate data and respiratory rate data are sent to a central processor integrated in the smart wearable underwear.

3. According to claim 1, a real-time monitoring method of vital signs based on smart wearable underwear is characterized in that: The user's current state is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the user's vital sign data threshold is obtained based on the user's current state, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period, including: The user's current activity state is obtained through the built-in accelerometer and gyroscope of the smart wearable underwear, and the activity state includes: stillness, walking, jogging and sprinting; The conductivity data of the user's skin is obtained through the built-in skin conductivity sensor of the smart wearable underwear; Obtain the respiratory rate and heart rate data of the same historical time period by integrating EEG sensors at the collar and shoulders of the smart wearable underwear to obtain the amplitude of the user's EEG signal; Determine the user's sleeping posture through the built-in accelerometer of the smart wearable underwear; The user's vital sign thresholds are obtained based on the vital sign threshold model.

4. According to claim 3, a real-time monitoring method of vital signs based on smart wearable underwear is characterized in that: The user's vital sign thresholds are obtained based on the vital sign threshold model, including: Acquire a user's heart rate threshold and a breathing rate threshold based on a heart rate threshold model and a breathing rate threshold model; Specifically, the heart rate threshold model is: ; in, represents the user's heart rate threshold at time t, , and represents the weight coefficient, + + =1, n represents the time window length for calculating the current heart rate trend, represents the heart rate at time i, m represents the number of time periods considering the heart rate of the same time period in the past, represents the time period weight coefficient, represents the heart rate in the same period in the past, ΔS represents the change in skin conductivity at time t relative to the previous moment, represents the standard deviation of skin conductivity variation, represents the user's basic heart rate, β represents the activity influence coefficient, represents the weight of each activity type, represents the proportion of each activity type at time t, and s represents the number of activity types; The respiratory rate threshold model is: ； in, represents the user's breathing rate threshold at time t, , and represents the weight coefficient, + + =1, represents the respiratory frequency at time i, p represents the time window length for calculating the current respiratory frequency trend, q represents the number of cycles of respiratory frequency considering the same sleep cycle in the past, represents the period weight coefficient, Indicates the breathing rate of the same sleep cycle in the past. Indicates the amplitude change of the EEG signal at time t relative to the previous time. Indicates the standard deviation of the EEG signal amplitude change, Indicates the user's basic breathing rate, represents the influence coefficient of sleeping posture, represents the weight of each sleeping posture, It represents the proportion of each sleeping posture at time t, and r represents the number of sleeping posture types, including supine, side-lying and prone.

5. According to claim 1, a real-time monitoring method of vital signs based on smart wearable underwear is characterized in that: When the user's vital sign data exceeds the threshold, an early warning is issued, the abnormal vital sign data of the user who issued the early warning is obtained, and an abnormal prediction model is trained based on the abnormal vital sign data of the user, including: When the user's vital sign data exceeds a threshold, an early warning is issued, and at the same time, abnormal vital sign data of the user within a time window when the early warning is issued is obtained, and features in the abnormal vital sign data are extracted, wherein the features include: average heart rate, maximum heart rate, minimum heart rate, rising slope of heart rate, falling slope of heart rate, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle, and standard deviation of respiratory rate within a respiratory cycle; The abnormal vital sign data are divided into a training set and a test set, and the long short-term memory network model is trained using the data in the training set, and the parameters are adjusted to the optimal value; Receive new vital sign data in real time, input the vital sign data into the pre-trained model, and the long short-term memory network model predicts whether the vital sign data has abnormal conditions based on the input real-time data, and issues an early warning if an abnormality is predicted.

6. A real-time vital sign monitoring system based on smart wearable underwear, characterized in that: The system comprises: A module for acquiring vital signs data, which acquires the user's vital signs data based on sensors integrated in the smart wearable underwear; A module for obtaining a vital sign data threshold value is used to obtain the user's current motion activity state through the accelerometer and gyroscope built into the smart wearable underwear, and obtain the user's vital sign data threshold value based on the user's current motion activity state, the user's vital sign data changes within a preset time window, and the user's historical vital sign data in the same time period; The abnormality prediction module is used to issue an early warning when the user's vital signs data exceeds a threshold, obtain the abnormal vital signs data of the user for issuing the early warning, train an abnormality prediction model based on the abnormal vital signs data of the user, and predict the abnormal situation of the user's vital signs based on the abnormality prediction model.

7. According to claim 6, a real-time vital sign monitoring system based on smart wearable underwear is characterized in that: The module for acquiring vital signs data includes: A heart rate data collection module, which is used for the flexible electrocardiogram sensor integrated in the chest position of the smart wearable underwear to collect the user's heart rate data according to a preset collection frequency; A respiratory frequency data collection module, which is used for collecting the user's respiratory frequency data according to a preset collection frequency through a piezoelectric sensor integrated in the abdomen of the smart wearable underwear; The sending module is used to send the collected heart rate data and respiratory rate data to a central processor integrated in the smart wearable underwear.

8. According to claim 6, a real-time vital sign monitoring system based on smart wearable underwear is characterized in that: The module for obtaining the threshold value of vital sign data includes: The activity status acquisition module is used to acquire the user's current activity status through the accelerometer and gyroscope built into the smart wearable underwear, and the activity status includes: stillness, walking, jogging and sprinting; A conductivity data acquisition module is used to acquire the conductivity data of the user's skin through the built-in skin electrical sensor of the smart wearable underwear; The module for obtaining the amplitude of the EEG signal is used to obtain the respiratory rate and heart rate data of the same historical time period by integrating EEG sensors at the collar and shoulders of the smart wearable underwear to obtain the amplitude of the user's EEG signal; The sleeping posture determination module is used to determine the user's sleeping posture through the accelerometer built into the smart wearable underwear; The vital sign threshold calculation module is used to obtain the user's vital sign threshold based on the vital sign threshold model.

9. The real-time vital sign monitoring system based on smart wearable underwear according to claim 8, characterized in that: The module for calculating the vital sign threshold value comprises: A heart rate threshold calculation module is used to obtain a user's heart rate threshold and a respiratory rate threshold based on a heart rate threshold model and a respiratory rate threshold model; Specifically, the heart rate threshold model is: ; in, represents the user's heart rate threshold at time t, , and represents the weight coefficient, + + =1, n represents the time window length for calculating the current heart rate trend, represents the heart rate at time i, m represents the number of time periods considering the heart rate of the same time period in the past, represents the time period weight coefficient, represents the heart rate in the same period in the past, ΔS represents the change in skin conductivity at time t relative to the previous moment, represents the standard deviation of skin conductivity variation, represents the user's basic heart rate, β represents the activity influence coefficient, represents the weight of each activity type, represents the proportion of each activity type at time t, and s represents the number of activity types; Calculate the respiratory rate threshold module. Specifically, the respiratory rate threshold model is: ； in, represents the user's breathing rate threshold at time t, , and represents the weight coefficient, + + =1, represents the respiratory frequency at time i, p represents the time window length for calculating the current respiratory frequency trend, q represents the number of cycles of respiratory frequency considering the same sleep cycle in the past, represents the period weight coefficient, Indicates the breathing rate during the same sleep cycle in the past. Indicates the amplitude change of the EEG signal at time t relative to the previous time. Indicates the standard deviation of the EEG signal amplitude change, Indicates the user's basic breathing rate, represents the influence coefficient of sleeping posture, represents the weight of each sleeping posture, It represents the proportion of each sleeping posture at time t, and r represents the number of sleeping posture types, including supine, side-lying and prone.

10. The real-time vital sign monitoring system based on smart wearable underwear according to claim 6, characterized in that: The abnormality prediction module includes: A feature extraction module is used to issue an early warning when the user's vital sign data exceeds a threshold value, and simultaneously obtain abnormal vital sign data of the user within a time window when the early warning is issued, and extract features from the abnormal vital sign data, wherein the features include: average heart rate, maximum heart rate, minimum heart rate, rising slope of heart rate, falling slope of heart rate, average respiratory rate, maximum respiratory rate, minimum respiratory rate, average respiratory cycle, and standard deviation of respiratory rate within a respiratory cycle; A training module, used to divide the abnormal vital sign data into a training set and a test set, use the data in the training set to train the long short-term memory network model, and adjust the parameters to the optimal level; The prediction module is used to receive new vital sign data in real time and input the vital sign data into the pre-trained model. The long short-term memory network model predicts whether there are any abnormalities in the vital sign data based on the input real-time data, and issues an early warning if an abnormality is predicted.

Images