Medical grade vsm dynamic monitoring system and method

Abstract

The application discloses a medical-grade VSM dynamic monitoring system and method, and relates to the technical field of vital sign monitoring. The system comprises a data processing module connected with an inertial measurement module and a display and interaction module; a power management module connected with the inertial measurement module, the data processing module and the display and interaction module; a display and interaction module used for obtaining basic information of a user to be monitored; a sensing module used for measuring vital sign signals and sending the vital sign signals to the data processing module; an inertial measurement module used for measuring inertial signals and sending the inertial signals to the data processing module; and the data processing module used for obtaining vital sign parameters of the user to be monitored according to the inertial signals and the vital sign signals, obtaining vital sign standard parameters according to the basic information and the inertial signals, comparing the vital sign parameters with the vital sign standard parameters to generate a monitoring report, and sending the monitoring report to the display and interaction module to output the monitoring report. The application can protect the safety of personal information of a user.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vital sign monitoring, in particular to a medical VSM dynamic monitoring system and method. BACKGROUND

[0002] With the development of society and the continuous improvement of people's health awareness, the demand for medical health services is increasing and more diversified.

[0003] Traditional vital sign monitoring mainly relies on patients going to medical institutions for manual measurement by professional medical staff at a specific time and place. However, for patients with chronic diseases who need long-term monitoring of vital signs, frequent visits to the hospital not only consumes time and energy, but also may affect the timeliness and continuity of monitoring due to factors such as traffic and queuing.

[0004] With the advent of various miniaturized, high-precision and wearable sensors, as well as the development of communication technology, vital sign remote monitoring technology has emerged, which enables patients to continuously monitor their vital signs in daily life scenarios and then send them to the medical information management system set in the hospital. Medical staff can remotely and real-time access the vital sign data of patients through the medical information management system, thereby realizing remote diagnosis, treatment plan adjustment and health guidance services.

[0005] However, in the process of transmitting the vital sign data of users, vital sign remote monitoring technology faces the risk of being hacked and leaked, and once a data leakage event occurs, it will cause great damage to the privacy of patients. SUMMARY

[0006] The purpose of the present application is to provide a medical VSM dynamic monitoring system and method that can protect the security of personal information of users.

[0007] To achieve the above purpose, the present application provides the following solutions:

[0008] In a first aspect, the present application provides a medical VSM dynamic monitoring system, which comprises a sensing module, an inertial measurement module, a data processing module, a display and interaction module, a warning prompt module and a power management module.

[0009] The data processing module is connected with the inertial measurement module and the display and interaction module respectively; the power management module is connected with the inertial measurement module, the data processing module and the display and interaction module respectively.

[0010] The display and interaction module is configured to acquire basic information of a user to be monitored, the basic information including age, gender, height, weight, altitude, and a monitoring mode, the monitoring mode including automatic monitoring or selective targeted monitoring.

[0011] The sensing module is configured to measure vital sign signals and send the vital sign signals to the data processing module.

[0012] The inertial measurement module is configured to measure inertial signals and send the inertial signals to the data processing module.

[0013] The data processing module is configured to acquire vital sign parameters of the user to be monitored according to the inertial signals and the vital sign signals, acquire vital sign standard parameters according to the basic information and the inertial signals, compare the vital sign parameters with the vital sign standard parameters, generate a monitoring report according to a comparison result, and send the monitoring report to the display and interaction module.

[0014] The display and interaction module is further configured to output the monitoring report.

[0015] In a second aspect, the present application provides a medical-grade VSM dynamic monitoring method, which is applied to the medical-grade VSM dynamic monitoring system as described in the first aspect, the medical-grade VSM dynamic monitoring system including a sensing module, an inertial measurement module, a data processing module, a display and interaction module, a pre-warning prompt module, and a power management module, the data processing module being connected with the inertial measurement module and the display and interaction module respectively, the power management module being connected with the inertial measurement module, the data processing module, and the display and interaction module respectively, and the method including:

[0016] The display and interaction module acquires basic information of a user to be monitored, the basic information including age, gender, height, weight, altitude, and a monitoring mode, the monitoring mode including automatic monitoring or selective targeted monitoring.

[0017] The sensing module measures vital sign signals and sends the vital sign signals to the data processing module.

[0018] The inertial measurement module measures inertial signals and sends the inertial signals to the data processing module.

[0019] The data processing module acquires a vital sign parameter of the user to be monitored according to the inertial signal and the vital sign signal, acquires a vital sign standard parameter according to the basic information and the inertial signal, compares the vital sign parameter with the vital sign standard parameter, generates a monitoring report according to the comparison result, and sends the monitoring report to the display and interaction module;

[0020] The display and interaction module outputs the monitoring report.

[0021] According to the specific embodiments provided in the application, the following technical effects are disclosed:

[0022] The application provides a medical-grade VSM dynamic monitoring system and method. The medical-grade VSM dynamic monitoring system comprises a sensing module, an inertial measurement module, a data processing module, a display and interaction module, and a power management module. The data processing module is connected with the sensing module, the inertial measurement module, and the display and interaction module. The power management module is connected with the inertial measurement module, the data processing module, and the display and interaction module. The display and interaction module is used to acquire basic information of a user to be monitored, including age, gender, height, weight, altitude, and a monitoring mode. The monitoring mode includes automatic monitoring or selective targeted monitoring. The sensing module is used to measure a vital sign signal and send the vital sign signal to the data processing module. The inertial measurement module is used to measure an inertial signal and send the inertial signal to the data processing module. The data processing module is used to acquire a vital sign parameter of the user to be monitored according to the inertial signal and the vital sign signal, acquire a vital sign standard parameter according to the basic information and the inertial signal, compare the vital sign parameter with the vital sign standard parameter, generate a monitoring report according to the comparison result, and send the monitoring report to the display and interaction module. The display and interaction module is also used to output the monitoring report. The vital sign signal can be acquired by the local sensing module, and the inertial signal can be acquired by the local inertial measurement module. The acquired vital sign signal and inertial signal are sent to the local data acquisition module. Thus, the data acquisition module can acquire the vital sign parameter of the user to be monitored based on the vital sign signal and the inertial signal, compare the vital sign parameter with the vital sign standard parameter, and generate a monitoring report output to the user locally. Since the relevant signal does not need to be sent to the remote end for generating the monitoring report through the network, the risk of hacking and leakage of the information of the user to be monitored in the network transmission process is avoided, the privacy of the patient is protected, and the security of the personal information of the user is improved. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] Figure 1 is a schematic diagram of a medical-grade VSM dynamic monitoring system according to an exemplary embodiment;

[0025] Figure 2 is a schematic diagram of a medical-grade VSM dynamic monitoring system according to an exemplary embodiment;

[0026] Figure 3 is a schematic diagram of a medical-grade VSM dynamic monitoring system according to an exemplary embodiment.

[0027] Legend of reference signs:

[0028] 1-sensor module; 11-wearable sensor; 12-analog front end; 2-inertial measurement module; 3-data processing module; 4-display and interaction module; 5-power management module; 6-SOS rescue module; 7-early warning prompt module. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor are within the scope of protection of the present application.

[0030] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Current healthcare or monitoring systems typically rely on external networks or cloud-based solutions to collect and analyze vital signs such as heart rate, respiratory rate, and blood oxygen saturation.

[0032] However, this kind of solution has the following defects:

[0033] 1. These systems usually rely on solutions connected through Wi-Fi, Bluetooth or cellular networks, which may cause potential problems in data privacy, security and dependence on Internet access, such as:

[0034] Reliability issues: Network interruptions or transmission delays can affect the real-time performance and reliability of the system.

[0035] Data security risks: During network transmission, data may be stolen, resulting in the risk of unauthorized access or leakage and tampering.

[0036] Limited applicability: In a network-free environment (such as remote areas, the wild, or the battlefield), it cannot work normally.

[0037] 2. In many cases, such as mountain climbing, ambulance service, or home care, the effectiveness of health monitoring systems may be limited by network environments, lack of reliable Internet access, or cloud infrastructure.

[0038] 3. High cost: These systems rely on network infrastructure and associated software systems, increasing the overall cost of use.

[0039] 4. Remote monitoring defects: These systems require external intervention and are limited in timeliness by network efficiency, especially in early warning or emergency measures that are not in the hands of the wearer themselves, but in the hands of third-party services, and the efficiency and quality problems of third-party services may affect the patient's missed treatment.

[0040] 5. The warning information of these systems is often directly transmitted to medical personnel, and the patient cannot see the monitoring data.

[0041] 6. These systems are often based on non-standard lead wearable medical solutions, such as watches, which can be called medical solutions, but are not medical-grade monitoring solutions. Medical-grade vital sign monitoring solutions, such as electrocardiogram monitors, are based on international AHA and IEC lead standards.

[0042] In summary, it is of great significance to develop a completely local solution (without connecting to external networks) that completely operates independently of the network, a vital sign monitoring device that can meet medical-grade monitoring and adapt to the needs of various scenarios.

[0043] Figure 1 is a schematic diagram of a medical-grade VSM dynamic monitoring system according to an example embodiment, as shown in Figure 1 The medical-grade VSM dynamic monitoring system includes a sensing module 1, an inertial measurement module 2, a data processing module 3, a display and interaction module 4, and a power management module 5.

[0044] The data processing module 3 is connected to the sensing module 1, the inertial measurement module 2, and the display and interaction module 4, respectively.

[0045] The power management module 5 is connected with the inertial measurement module 2, the data processing module 3 and the display and interaction module 4 respectively; the power management module 5 in the present disclosure supports long endurance time, for example, satisfies 24 hours to 15 days, even 30 days or longer time of vital sign signal monitoring, for example: can be equipped with cochlea battery, rechargeable high-energy battery, micro nuclear battery and the like. In order to further prolong the working time, the power management module 5 of the present disclosure is also built-in low-power circuit design.

[0046] The display and interaction module 4 is used for acquiring basic information of the user to be monitored, and the basic information includes age, gender, height, weight, altitude and monitoring mode; the monitoring mode includes automatic monitoring or selective targeted monitoring.

[0047] The sensing module 1 is used for measuring the vital sign signal of the user to be monitored and sending the vital sign signal to the data processing module 3; wherein the vital sign signal includes at least one of the following signals: ECG signal, PPG signal, BIOZ signal and temperature signal.

[0048] The inertial measurement module 2 is used for measuring the inertial signal of the user to be monitored and sending the inertial signal to the data processing module 3; wherein the inertial measurement module 2 can include: three-axis accelerometer, three-axis gyroscope and three-axis magnetometer, the three kinds of devices can be used independently (such as only selecting three-axis accelerometer), or can be used in combination, and the present disclosure is described by taking three-axis accelerometer as an example.

[0049] The data processing module 3 is configured to obtain a vital sign parameter of the user to be monitored according to the inertial signal and the vital sign signal, obtain a vital sign standard parameter according to the basic information and the inertial signal, compare the vital sign parameter with the vital sign standard parameter, generate a monitoring report according to a comparison result, and send the monitoring report to the display and interaction module 4. The data processing module 3 in the present disclosure uses a high-performance microcontroller unit, including but not limited to a floating-point operation-based FPU, an ultra-low-power artificial intelligence (AI), an MCU (such as a convolutional neural microprocessor CNN-MCU), a recurrent neural RNN-MCU, and the like. The data processing module 3 has the following functions: real-time acquisition, filtering, data analysis, feature extraction, optimization, and judgment of the vital sign signal, and output of a warning prompt module 7. The CNN-MCU has an AI self-adaptive learning vital sign analysis function. The data is stored and displayed locally. Based on the convolutional neural data, a local diagnosis or auxiliary diagnosis report is generated (without displaying an electrocardiogram, and the internal processing result of the electrocardiogram analysis is partially displayed). The vital sign parameter (Vital Sign Parameters, referred to as VSM) includes at least one of the following parameters: heart rate (HR), heart rate variability (HRV), respiratory rate (RR), blood oxygen saturation (SpO2), blood pressure (BP), arrhythmia (Rhythm), and body temperature (T). The arrhythmia includes atrial fibrillation (AF), premature ventricular contraction (PVC), and wet tachycardia (VT).

[0050] In the process of obtaining the vital sign standard parameter according to the basic information and the inertial signal, the motion state can be obtained based on the inertial signal first, and then the HR, the RR, the SpO2, the BP, and the HRV in the vital sign standard parameter can be determined based on the motion state and the age and the gender in the basic information. Then, the body temperature reference standard in the vital sign standard parameter can be obtained based on the age.

[0051] Taking the acceleration as an example, after obtaining the acceleration of the user to be monitored, the motion state of the user to be monitored is obtained based on Table 1.

[0052] Table 1

[0053]

[0054] The HR, the RR, the SpO2, the BP, and the HRV in the vital sign standard parameter are determined based on the motion state and the age and the gender in the basic information, as shown in Table 2.

[0055] Table 2

[0056]

[0057]

[0058]

[0059] It is worth noting that in determining the HR, RR, SpO2, BP and HRV in the vital sign standard parameters, attention can also be paid to the current season, specifically:

[0060] HR: 5-10% higher in winter than in summer;

[0061] HRV: 10-15% lower in winter than in summer;

[0062] RR: 5-8% higher in winter than in summer;

[0063] SpO2: 1-2% lower in winter than in summer (cold causes blood vessels to constrict).

[0064] Based on age, the reference standard for body temperature in the vital sign standard parameters is obtained, as shown in Table 3:

[0065] Table 3

[0066]

[0067]

[0068] In clinical practice, normal body temperature is generally between 36.5°C and 37.5°C, but the normal temperature range for each person may be slightly different. And with age, body temperature may decrease slightly, especially in the super-elderly. Body temperature also varies depending on time (such as morning and evening), activity level, diet, and other factors.

[0069] Body temperature measurement methods: Common methods for measuring body temperature include oral, underarm, rectal, ear, and forehead. Different measurement methods have different normal ranges. For example, the body temperature measured under the armpit is usually slightly lower than the body temperature measured rectally.

[0070] Generally, a body temperature above 38°C is considered to be fever, and a body temperature above 39°C may indicate a more serious infection or other disease.

[0071] The display and interaction module 4 is also used to output the monitoring report. In the present disclosure, the display and interaction module 4 can integrate a high-resolution screen (including but not limited to, such as LCD, OLED, TFT, etc.), display the monitoring report, or can also inherit a touch screen or physical button operation, facilitate user input. The display and interaction module 4 can also provide a human-machine interface (Human-Machine Interface, abbreviated as: HMI), which can access a human-machine voice interaction device, etc., to facilitate user operation.

[0072] The medical-grade VSM dynamic monitoring system in the present disclosure can complete monitoring and display of monitoring reports locally without the need for transmission to other device ends through a network, thereby avoiding the problem of network interruption or transmission delay affecting the real-time performance and reliability of the system, and can avoid the risk of unauthorized access, leakage and tampering of data during transmission, and can operate normally even in a network-free environment, thereby expanding the scope of use.

[0073] The medical-grade VSM dynamic monitoring system in the present disclosure does not rely on network infrastructure and associated software systems, thereby reducing the use cost.

[0074] By using the medical-grade VSM dynamic monitoring system in the present disclosure, the patient himself can directly see the monitoring data.

[0075] The medical-grade VSM dynamic monitoring system in the present disclosure provides a reliable, private and autonomous continuous health monitoring method for a completely local medical-grade VSM dynamic monitoring solution without the need for external networks or devices. The design focuses on real-time data analysis, alerts, privacy and local storage, ensuring that users can obtain accurate health information and immediate feedback regardless of the environment. By addressing key challenges in remote monitoring, patient data privacy and healthcare, the present disclosure fills a key gap in the current offline vital sign monitoring system market. It meets medical standards (such as IEC, ISO and FDA), further consolidating its applicability in various healthcare environments. With its flexibility, future scalability and compliance, the medical-grade VSM dynamic monitoring system will become an important tool for personal health management, emergency care and remote patient monitoring.

[0076] In one embodiment, as shown in FIG. 1, the sensor module 1 comprises a wearable sensor 11 and an analog front end 12. Figure 2

[0077] The analog front end 12 is connected with the wearable sensor 11, the data processing module 3 and the power management module 5 respectively.

[0078] The wearable sensor 11 is configured to collect raw vital sign signals and send the collected raw vital sign signals to the analog front end 12.

[0079] The analog front end 12 is configured to pre-process the received raw vital sign signals to obtain vital sign signals.

[0080] The sensor module 1 is composed of the wearable sensor 11 and the analog front end 12 (Analog Front-End, referred to as AFE for short).

[0081] The wearable sensor 11 can include a wearable garment, a patch lead electrode, an optical sensor, a digital temperature sensor or other temperature sensors, etc.​

[0082] Analog Front End 12 AFE, preferably an advanced clinical medical grade AFE, AFE for vital sign parameter analog-digital conversion, etc. Such as integrated ECG, BIOZ, PPG sensor, can easily obtain ECG, PPG, BIOZ signal, used to extract vital sign parameters (VSM).

[0083] Among them, the AC dynamic input range of the amplifier built by AFE is ± 50mVP-p~± 200mVP-p.

[0084] In this disclosure, medical grade AFE is adopted, which can achieve medical grade monitoring.

[0085] In one embodiment, as shown in Figure 3 The medical grade VSM dynamic monitoring system in the present disclosure further comprises an SOS rescue module 6.

[0086] The SOS rescue module 6 is connected with the data processing module 3 and the power management module 5 respectively.

[0087] The data processing module 3 is further configured to generate a help signal when an abnormal vital sign parameter is detected, and send the help signal to the SOS rescue module 6.

[0088] The SOS rescue module 6 is configured to execute a preset help means after receiving the help signal, and the preset help means comprises sending a help information to a preset emergency contact and / or sending a help information to a rescue agency.

[0089] When the SOS rescue module 6 sends the help information, it can also carry the patient's location, vital sign data and other key information.

[0090] By setting the SOS rescue module 6, when the patient has an abnormal vital sign or encounters an emergency, the SOS rescue module 6 can quickly send a help signal. This allows medical staff or related rescue personnel to know the critical condition of the patient in the first time, greatly shortening the rescue time and giving the patient more rescue opportunities.

[0091] In one embodiment, the data processing module 3 is further configured to:

[0092] Obtain medical advice according to the abnormal vital sign parameter, and the medical advice provides measures for the user to be monitored to cope with the abnormal vital sign parameter.

[0093] The monitoring report of the present disclosure comprises:

[0094] Vital sign parameters, abnormal vital sign parameters and doctor's advice.

[0095] For example, the monitoring report usually includes the following contents:

[0096] Vital sign parameter statistics: statistical data of HR, RR, SpO2, BP, HRV, T, etc. parameters, such as: mean, maximum, minimum, predicted trend, etc.

[0097] Fluctuation range of each index.

[0098] Abnormality detection: record abnormal events, mark the time of occurrence and specific abnormal conditions. For example: identify atrial fibrillation (AF), premature ventricular contraction (PVC) or ventricular tachycardia (VT).

[0099] Provide early warning information, mark the triggered early warning threshold.

[0100] Medical advice: generate corresponding medical advice (such as suggesting hospitalization, stopping strenuous exercise, etc.) according to the monitoring results.

[0101] The early warning mode of the monitoring and early warning of the present disclosure can include: visual, auditory, tactile, etc.

[0102] The reporting mode of the monitoring report of the present disclosure can include: visual and auditory modes, wherein the visual can include display through the display screen, and the auditory mode can include voice, etc.

[0103] a) Vital sign parameter statistics according to Table 2.

[0104] b) Abnormality detection:

[0105] · No atrial fibrillation (AF) occurred.

[0106] · No premature ventricular contraction (PVC) occurred.

[0107] · No ventricular tachycardia (VT) occurred.

[0108] c) Medical advice:

[0109] · The results of this monitoring are good, no abnormalities, continue to maintain a healthy lifestyle.

[0110] · If the symptoms change, please contact the doctor in time.

[0111] In this way, you can ensure that in the VSM system, the real-time monitoring and early warning system can effectively find potential health risks, generate monitoring reports in time, and provide necessary medical intervention suggestions.

[0112] In one embodiment, the data processing module 3 is specifically configured to perform the following steps A1-A4:

[0113] A1, obtain the body mass index (BMI) of the user to be monitored according to the height and weight, which can be obtained by the following formula:

[0114]

[0115] Wherein, the weight unit is Kg, and the height unit is m.

[0116] Table 4 shows the BMI and health risk classification:

[0117] Table 4

[0118] Classification BMI range (kg / m 2 )]]> Health Risk Underweight <18.5 Increased risk of malnutrition-related diseases Normal weight 18.5–24.9 Lowest health risk Overweight class I (pre-obese) 25.0–29.9 Slightly increased health risk Overweight class II (obese class I) 30.0–34.9 Moderately increased health risk Overweight class III (obese class II) 35.0–39.9 Severely increased health risk Overweight class IV (extremely obese class III) ≥40.0 Very high health risk

[0119] A2, obtain the body mass index coefficient and the altitude coefficient according to the target basic parameter; the target basic parameter includes: the body mass index, or the altitude.

[0120] The body mass index coefficient and the altitude coefficient can be obtained based on Table 5:

[0121] Table 5

[0122] Classification Body mass index coefficient C bmi ]] Altitude coefficient C altitud ]]> Low weight (BMI < 18.5) About 0.5 About 1.0 Normal weight (18.5 < BMI < 24.9) About 1.0 About 1.0 Overweight (25 < BMI < 29.9) About 1.5-2.0 About 1.1-1.3 Obese (BMI > 30) About 2.0-3.0 About 1.3-2.0 Low altitude (0-1000 meters) About 1.0 About 1.0 Medium altitude (1000-3000 meters) About 1.0 About 1.1-1.3 High altitude (3000 meters and above) About 1.0 About 1.3-2.0

[0123] After obtaining the body mass index coefficient and the altitude coefficient, the low altitude can be used as the demarcation line, when the altitude is low, the body mass index coefficient and the altitude coefficient corresponding to the BMI classification are used; when the altitude is medium or high, the body mass index coefficient and the altitude coefficient corresponding to the altitude classification are used.

[0124] A3, optimize the inertial signal according to the body mass index coefficient and the altitude coefficient.

[0125] Continue to take the inertial signal as an example of the acceleration signal:

[0126] The acceleration signal is collected by a three-axis accelerometer (such as ADXL367), which usually provides acceleration components in X, Y, Z three directions, with units of g or m / s 2 , for example, the three-axis accelerometer provides: a x (t), a y (t), a z (t), wherein a x (t) is the acceleration component on the X axis, a y (t) is the acceleration component on the Y axis, and a z (t) is the acceleration component on the Z axis, which changes with time t.

[0127] The total acceleration a(t) of the device is calculated by the three-axis acceleration components:

[0128]

[0129] wherein a(t) has a unit of g or m / s 2 , represents the motion intensity or motion state of the triaxial accelerometer in three-dimensional space. The acceleration is optimized using a body mass index coefficient and an altitude coefficient, and the formula is as follows:

[0130] A(t) = C bmi ·C altiud ·a(t);

[0131] wherein C bmi is a body mass index coefficient, and C altitud is an altitude coefficient.

[0132] A4, determining the vital sign parameters of the user to be monitored according to the optimized inertial signal and the vital sign signal.

[0133] In one embodiment, in terms of determining the vital sign parameters of the user to be monitored according to the optimized inertial signal and the vital sign signal, the data processing module 3 is specifically configured to perform the following steps A41-A42:

[0134] A41, obtaining filter parameters according to the optimized inertial signal, and processing the vital sign signal according to the filter parameters to obtain a target vital sign signal.

[0135] For example, filter parameters can be obtained according to the optimized inertial signal, and the vital sign signal can be filtered according to the filter parameters, at this time, the vital sign signal after filtering is the target vital sign signal.

[0136] Filter parameters can also be obtained according to the optimized inertial signal, and the vital sign signal can be filtered according to the filter parameters; then the vital sign signal after filtering is smoothed, at this time, the vital sign signal after smoothing is the target vital sign signal.

[0137] In one embodiment, in terms of obtaining filter parameters according to the optimized inertial signal, the data processing module 3 is specifically configured to:

[0138] obtain the dynamic gain of the high-pass filter and the cutoff frequency of the high-pass filter according to the optimized inertial signal, filter the vital sign signal according to the dynamic gain of the high-pass filter and the cutoff frequency of the high-pass filter, and obtain a high-pass filtered signal of the vital sign signal;

[0139] Specifically: according to the optimized acceleration A(t), the threshold or motion threshold A thresh (unit g or m / s 2) and gain adjustment coefficient β, dynamically calculate the gain G of the high-pass filter highpass (t) and the cutoff frequency f of the high-pass filter highpass .

[0140] The calculation formula of the optimized acceleration A(t) and the gain of the high-pass filter is:

[0141]

[0142] Wherein, the gain G0 of the high-pass filter represents the initial gain of the high-pass filter, which is usually set to 1, A thresh represents the threshold of the motion intensity when calculating the gain of the high-pass filter, which determines the division of the motion state, β represents the adjustment coefficient of the gain of the high-pass filter, which controls the influence intensity of the optimized acceleration on the gain of the high-pass filter, usually β>1.

[0143] Specifically, in addition to gain adjustment, the cutoff frequency of the high-pass filter can also be dynamically adjusted according to the acceleration. Generally, when the motion intensity increases, the cutoff frequency of the high-pass filter can be increased to better remove motion artifacts. For example, assuming the cutoff frequency f highpass is dynamically adjusted with the change of the optimized acceleration A(t):

[0144]

[0145] Wherein, f highpass represents the cutoff frequency of the high-pass filter, f0 represents the initial cutoff frequency of the high-pass filter, which is set to a proper value (such as 10HZ), δ represents the adjustment coefficient of the cutoff frequency of the high-pass filter, A(t) represents the influence degree of the increase of the optimized acceleration on the cutoff frequency of the high-pass filter, A thresh represents the threshold of the motion intensity.

[0146] Wherein, when the motion intensity increases, the cutoff frequency f highpass of the high-pass filter will increase, so that it can more effectively remove low-frequency interference.

[0147] And / or, according to the optimized inertial signal, the dynamic gain of the low-pass filter and the cutoff frequency f lowpass of the low-pass filter are obtained, and the vital sign signal is filtered according to the dynamic gain of the low-pass filter and the cutoff frequency of the low-pass filter to obtain the low-pass filtered signal of the vital sign signal.

[0148] Specifically: according to the optimized acceleration A(t), through the pre-set threshold or motion threshold A t ′ hresh (unit g or m / s 2 ) and adjustment coefficient, dynamically calculate the gain G of the low-pass filterlowpass (t) and the cutoff frequency f of the low-pass filter lowpass .

[0149] The calculation formula of the low-pass filter gain obtained by the optimized acceleration A(t) is:

[0150]

[0151] wherein G lowpass (t) represents the gain of the low-pass filter, represents the initial gain of the low-pass filter, which is usually set to 1, represents the threshold of the motion intensity when calculating the gain of the low-pass filter, which determines the division of the motion state, a represents the gain adjustment coefficient of the low-pass filter, which controls the influence degree of the motion state on the gain of the low-pass filter, and is usually set to 0≤a≤1, and γ represents the gain adjustment coefficient of the low-pass filter, which controls the influence intensity of the optimized acceleration on the gain of the high-pass filter, and is usually γ>0.

[0152] Specifically, in addition to the gain adjustment, the cutoff frequency of the low-pass filter can also be dynamically adjusted according to the acceleration. Generally, when the motion intensity decreases, the cutoff frequency of the low-pass filter can be reduced to better remove the motion artifacts. For example, assuming that the cutoff frequency f lowpass is dynamically adjusted with the change of the optimized acceleration A(t):

[0153]

[0154] wherein f lowpass represents the cutoff frequency of the low-pass filter, f'0 represents the initial cutoff frequency of the low-pass filter, which is set to a proper value (such as 10 Hz), δ' represents the adjustment coefficient of the cutoff frequency of the low-pass filter, A(t) represents the influence degree of the increase of the optimized acceleration on the cutoff frequency of the low-pass filter, and A' thresh represents the threshold of the motion intensity when calculating the gain of the low-pass filter.

[0155] wherein when the motion intensity decreases, the cutoff frequency f lowpass of the low-pass filter will be reduced, so that it can more effectively remove the high-frequency interference.

[0156] And / or, the dynamic gain of the band-pass filter and the cutoff frequency of the band-pass filter are obtained according to the optimized inertial signal, the vital sign signal is filtered according to the dynamic gain of the band-pass filter and the cutoff frequency of the band-pass filter, and a band-pass filtered signal of the vital sign signal is obtained.

[0157] Specifically: the calculation formula of the dynamic gain of the band-pass filter obtained by the optimized inertial signal is:

[0158] G bandpass (t) = Inverse FFT(A filtered (f));

[0159] A filtered (f) = A(f) · H(f);

[0160]

[0161] A(f) = FFT(A(t));

[0162] where G bandpass (t) is the dynamic gain of the band-pass filter, A filtered (f) is the output function of the acceleration signal through the band-pass filter; A(f) is the frequency spectrum obtained by FFT change on the optimized acceleration A(t), H(f) represents the transfer function of the band-pass filter, f high is the high frequency cutoff point of the band-pass filter, and f low is the low frequency cutoff point of the band-pass filter.

[0163] In one embodiment, in the aspect of processing the vital sign signal according to the filter parameters to obtain the target vital sign signal, the data processing module 3 is specifically used for:

[0164] The high-pass filter allows signals with a frequency higher than the cutoff frequency f highpass (t) to pass through, and for the vital sign signal X(t), the output X highpass (t) after high-pass filtering can be represented as:

[0165] X highpass (t) = G highpass (t) · X(t) · H highpass (f, f highpass (t));

[0166] where G highpass (t) represents the gain of the high-pass filter, H highpass (f, f highpass (t)) is the frequency response function of the high-pass filter, which is usually represented as a filter characteristic related to the cutoff frequency f highpass (t) of the high-pass filter.

[0167] The low-pass filter allows signals with a frequency lower than the cutoff frequency f lowpass (t) to pass through, and for the vital sign signal X(t), the output X lowpass (t) after low-pass filtering can be represented as:

[0168] X lowpass (t) = G lowpass(t) · X(t) · H lowpass (f, f lowpass (t));

[0169] wherein G lowpass (t) represents the gain of the low-pass filter, H lowpass (f, f lowpass (t)) is the frequency response function of the low-pass filter, which is usually represented as a filter characteristic related to the cut-off frequency f lowpass (t) of the low-pass filter.

[0170] A band-pass filter allows signals within a certain frequency range (determined by the low cut-off f low (t) and the high cut-off f high (t)) to pass, and the output X bandpass (t) of the band-pass filter after filtering can be represented as:

[0171] X bandpass (t) = G bandpass (t) * X(t) * H bandpass (f, f low (t), f high (t));

[0172] wherein G bandpass (t) is the dynamic gain of the band-pass filter, H bandpass (f, f low (t), f high (t)) is the frequency response function of the band-pass filter, which is usually represented as a filter characteristic related to the low cut-off f low (t) and the high cut-off f high (t).

[0173] That is, the high-pass filter mainly functions to filter out signals below a certain cut-off frequency and retain high-frequency components; the low-pass filter mainly functions to filter out signals above a certain cut-off frequency and retain low-frequency components; the band-pass filter allows signals within a specific frequency range to pass and removes components below the lowest cut-off frequency and above the highest cut-off frequency, and the frequency response function described above is usually determined by the specific way in which the filter is designed.

[0174] wherein the output X highpass (t) of the high-pass filter, and / or the output X lowpass (t) of the low-pass filter, and / or the output X bandpass (t) of the band-pass filter after filtering characterizes the target vital sign signal.

[0175] Further, in step A41, the filtered vital sign signal can be further smoothed by a smoothing algorithm (such as Moving Average, Weighted Moving Average, Exponential Smoothing) to further remove small fluctuations.

[0176] Taking the exponential smoothing algorithm as an example:

[0177] S(t) = ξ x m(t) + (1 - ξ) x S(t - 1);

[0178] where S(t) represents the smoothed signal value at time point t, m(t) represents the original value of the filtered vital sign signal at time t, S(t - 1) represents the smoothed signal value at time point (t - 1), and ξ represents a smoothing coefficient between 0 and 1, used to control the balance between new data and historical data.

[0179] A42, determining the vital sign parameter of the user to be monitored according to the target vital sign signal.

[0180] The acquisition process of each vital sign parameter is described in detail below.

[0181] I. Heart rate (HR) feature.

[0182] The target vital sign signal can be input into the trained target heart rate feature extraction convolutional neural network to obtain the predicted heart rate feature in the target vital sign signal through the trained target heart rate feature extraction convolutional neural network.

[0183] Before acquiring and predicting the heart rate feature, the initial heart rate feature extraction convolutional neural network needs to be trained to obtain the target heart rate feature extraction convolutional neural network. The specific training process includes the following steps B1-B6:

[0184] B1, obtaining a heart rate training sample set, the training sample set including: a training sample signal and a labeled label, wherein the labeled label can be understood as a labeled true heart rate.

[0185] B2, inputting the training sample signal into the initial heart rate feature extraction convolutional neural network to obtain a training heart rate prediction result;

[0186] Specifically, step B2 includes the following sub-steps B21-B24:

[0187] B21, inputting the training sample signal into the convolutional layer for convolution processing;

[0188] Among them, the convolution operation is the core part of extracting signal features in the convolutional neural network (CNN). It convolves the training sample signal X'(t) with the convolution kernel W, and the specific formula is:

[0189] F(t) = (X'(t) * W) + b;

[0190] Among them, F(t) represents the feature map after convolution, which represents the signal features obtained after convolution operation; W represents the convolution kernel, which represents a small filter, usually there are multiple different convolution kernels for extracting different types of features; * represents the convolution operator, which represents the convolution operation; b represents the bias term of the convolution layer, X'(t) represents the time domain data of the training sample signal, usually the signal in a time window.

[0191] B22, input the training sample signal after convolution processing into the pooling layer for pooling processing;

[0192] The pooling operation is used to reduce the dimension of the feature map and retain important features. The maximum pooling is a commonly used pooling method, which selects the maximum value in the local window, and the specific formula is:

[0193] F pool (t) = max(F(t));

[0194] Among them, F pool (t) represents the feature map after pooling, max() represents the maximum pooling operation, that is, selecting the maximum value in the pooling window.

[0195] B23, the training sample signal after pooling processing is analyzed in the frequency domain.

[0196] In order to extract the frequency features related to heart rate from the signal, fast Fourier transform (FFT) is often used for frequency domain analysis, and the specific formula is:

[0197] S(f) = FFT(F(t));

[0198] Among them, S(f) represents the frequency domain features of the signal F(t) after Fourier transform, which represents the intensity of the signal at different frequencies; f HR represents the main frequency related to heart rate, which represents the frequency component corresponding to heart rate in the frequency spectrum.

[0199] B24, input the training sample signal after frequency domain processing into the fully connected layer to obtain the training heart rate prediction result.

[0200] The fully connected layer integrates the features extracted by frequency domain analysis to output the final prediction result, and the specific formula is:

[0201] HR = FC(S(f), W FC ) + bFC ;

[0202] where W FC represents the weight matrix of the fully connected layer, b FC represents the bias term of the fully connected layer, and HR represents the initial training heart rate prediction result.

[0203] B25, output the final training heart rate prediction result through the output layer.

[0204] The output layer gives the final training heart rate prediction result, and the specific formula is as follows:

[0205]

[0206] where, The final training heart rate prediction result is shown, and f(HR) is a nonlinear activation function (such as ReLU, sigmoid, etc.), which is used to map the output of the model to the final training heart rate prediction result.

[0207] B3, input the training heart rate prediction result and the corresponding label into the loss function to obtain the loss function value.

[0208] In order to train the CNN model, it is necessary to define a loss function to measure the difference between the predicted value and the true value. The commonly used loss function is mean square error (MSE), especially in regression tasks, and the formula is as follows:

[0209]

[0210] where, represents the final training heart rate prediction result corresponding to the i-th training sample signal, HR i represents the true heart rate corresponding to the i-th training sample signal, and N represents the number of training sample signals.

[0211] B4, adjust the parameters of the initial heart rate feature extraction convolutional neural network according to the loss function value, to obtain the heart rate feature extraction convolutional neural network after parameter adjustment.

[0212] B5, quantization processing is performed on the heart rate feature extraction convolutional neural network after parameter adjustment.

[0213] Quantization mainly converts high-precision data (such as 32-bit floating-point numbers) into low-precision integer values, and the quantization formula can be:

[0214]

[0215] where χ represents the weight, activation value or input value of a certain layer in the heart rate feature extraction convolutional neural network after parameter adjustment; Δ represents the quantization step, which is usually determined by the precision of quantization, for example, when the step is 28 ; round(·) denotes rounding a value to the nearest integer.

[0216] This means that x will be scaled and rounded to the nearest low-precision representation, and after quantization, the model will become more efficient, reducing storage requirements and accelerating computation.

[0217] B6, the pruned heart rate feature extraction convolutional neural network is pruned to obtain the trained target heart rate feature extraction convolutional neural network.

[0218] Pruning is usually done by removing unimportant neurons or connections in the network to reduce computation. Assuming we have a neural network where the weight matrix of each layer is W, the pruning process can be defined as:

[0219] W pruned = W·L|w|>τ;

[0220] where W pruned represents the pruned weight matrix; |w| represents the absolute value of each element in the weight matrix w; τ represents the pruning threshold, and connections with weight values less than τ are removed; L represents the indicator function, and only when the absolute value of the weight is greater than the threshold τ, the connection will be retained.

[0221] Pruning is done by removing connections with small absolute weight values to reduce computational complexity and storage requirements.

[0222] It is worth noting that in the optimization method combining quantization and pruning, quantization and pruning are usually performed sequentially, or in some cases alternately. The goal of this method is to reduce storage requirements, computational complexity, and maintain model performance at the same time.

[0223] First, the pruning operation obtains the weight matrix W pruned ;

[0224] Then, the quantization operation:

[0225]

[0226] where Q(W pruned ) represents the pruned weight after quantization.

[0227] The combined operation can first prune unnecessary connections to reduce the number of connections, and then quantize to reduce the computational precision, thereby further improving efficiency.

[0228] Quantization reduces storage requirements and computation by mapping high-precision values to lower-precision integers.

[0229] Pruning reduces computational complexity and storage requirements by removing unimportant connections or neurons.

[0230] Quantization combined with pruning first reduces the network size through pruning, and then reduces the precision through quantization, thereby achieving the maximum optimization of resources.

[0231] The final HR value is the predicted value obtained through the output layer of the neural network. After filtering, convolution, pooling, frequency domain analysis, full connection, and other steps, the model outputs a predicted heart rate (HR). Through the loss function (such as MSE) in the training process, the model gradually adjusts the parameters so that the predicted HR value is as close to the true value as possible. After quantization and pruning optimization, the inference efficiency and accuracy of the model are improved, so that it can better predict heart rate in actual application.

[0232] In calculating the heart rate related features, the ECG signal or PPG signal in the target vital sign signal is used.

[0233] II. Heart rate variability (HRV) features.

[0234] Heart rate variability (HRV) feature extraction is based on the analysis of r-r intervals (also known as rr intervals) in the target vital sign signal, mainly including time domain, frequency domain and nonlinear analysis methods, which include the following steps C1-C3:

[0235] C1, obtain the R wave position in the target vital sign signal;

[0236] The R wave position in the target vital sign signal can be obtained through the R wave detection algorithm, and the commonly used ones are differential, square, and sliding window integration method, which detects the R wave position by setting the R wave threshold.

[0237] (1) Differentiation.

[0238] Differentiation can enhance the slope change of the signal and highlight the slope of the QRS complex, and the specific formula is:

[0239] Y diff [n]=X[n]-X[n-1];

[0240] Where X[n] and X[n-1] represent the input target vital sign signal, Y dif [n] represents the target vital sign signal after differentiation, highlighting the slope change.

[0241] (2) Squaring.

[0242] Squaring operation changes the signal to positive value, while enhancing the amplitude of QRS wave, and the specific formula is:

[0243] Y square [n]=(Tdiff [n] 2 ;

[0244] where Y Square [n] represents the squared target vital sign signal, used to enhance the detection of sharp waveforms.

[0245] (3) Moving Window Integration.

[0246] Moving Window Integration smooths the signal and emphasizes the overall features of the QRS waveform, with the specific formula:

[0247]

[0248] where Y int [n] represents the Moving Window Integrated target vital sign signal, N represents the length of the moving window, usually taking 150ms-200ms. k represents the offset of n to the past N time points (from k=0 to k=N-k=N-1).

[0249] After obtaining the processed target vital sign signal through the above three methods, a dynamic or static R-wave threshold is set to detect the R-wave position. The position where the signal value in the processed target vital sign signal is greater than the R-wave threshold is the R-wave position.

[0250] The R-wave position can also be determined by a QRS detection algorithm.

[0251] The purpose of the QRS detection algorithm is to identify a complete QRS waveform (including Q-wave, R-wave, S-wave), and its algorithm is based on signal feature extraction in a specific frequency range.

[0252] a) R-wave calibration.

[0253] The position of the R-wave is determined by finding the maximum value of the local signal:

[0254] R peak = max(x[n]) for n∈[t start , t end ];

[0255] where x[n] represents the input filtered target vital sign signal, R peak represents the position of the R-wave peak (time point n), and [t start , t end ] represents the time range of the moving window, usually covering one heart cycle.

[0256] b) Q-wave calibration.

[0257] The Q-wave is calibrated by finding the minimum value of the signal before the R-wave:

[0258] Q wave = min(x[n]) for n ∈ [t start , R peak ];

[0259] where Q wave denotes the position of Q-wave (time point n); [t start , R peak ] denotes the time range from the start point of the sliding window to the position of R-wave peak.

[0260] c) S-wave labeling.

[0261] Labeling S-wave by finding the minimum value of the signal after R-wave:

[0262] S wave = min(x[n]) for n ∈ [R peak , t end ];

[0263] where S wave denotes the position of S-wave (time point n); [Rpeak, t end ] denotes the time range from the position of R-wave peak to the end point of the sliding window.

[0264] C2, obtaining all the rr intervals in the target vital sign signal according to the positions of R-waves.

[0265] All the positions of R-waves are obtained through C1, and all the rr intervals can be obtained based on all the positions of R-waves.

[0266] C3, obtaining a heart rate variability feature value according to the rr intervals, the heart rate variability feature value including: an average heart rate interval, a heart rate, a standard deviation of all the rr intervals, a root mean square of adjacent rr interval differences, a proportion of rr interval differences greater than 50 ms, a power spectral density in a frequency band corresponding to each rr interval, a short-term HRV component, and a long-term HRV component.

[0267] Specifically, the HRV time domain feature extraction can adopt a time domain method to calculate the heart rate variability feature value by counting the rr intervals.

[0268] (1) Average heart rate interval.

[0269]

[0270] where M ean_rr denotes the average rr interval, with a unit of second; rr i denotes the i-th rr interval, and M denotes the total number of rr intervals.

[0271] (2) Heart rate.

[0272]

[0273] where HR represents heart rate, in bpm (beats per minute).

[0274] (3) Standard deviation.

[0275]

[0276] where SDNN represents the standard deviation of all rr intervals, reflecting the overall level of heart rate variability.

[0277] (4) Root mean square of successive differences.

[0278]

[0279] where RMSSD represents the root mean square of successive differences, mainly reflecting parasympathetic nervous activity; (rr i+1 -rr i ) represents the difference between adjacent rr intervals.

[0280] (5) Proportion of rr interval differences greater than 50 ms, pNN50:

[0281]

[0282] where pNN50 represents the proportion of rr interval differences greater than 50 ms, reflecting short-term fluctuations in heart rate variability.

[0283] Specifically, the HRV frequency domain feature extraction method calculates the power density of different frequency bands by performing spectral analysis on the rr interval sequence.

[0284] (1) Frequency band division.

[0285] ULF (ultra-low frequency): <0.003 Hz;

[0286] VLF (very low frequency): 0.003 Hz ≤ f < 0.04 Hz;

[0287] LF (low frequency): 0.04 Hz ≤ f < 0.15 HzH;

[0288] HF (high frequency): 0.15 Hz ≤ f < 0.4 Hz;

[0289] (2) Frequency spectrum power calculation

[0290] Use fast Fourier transform (FFT) or Lomb-Scargle spectrum analysis method to calculate the power of each frequency band:

[0291]

[0292] where P Band represents the power density in the frequency band, PSD(f) represents the power spectral density, and f1, f2 represent the start and end frequencies of the frequency band.

[0293] (3) Frequency domain features.

[0294] By comparing P Band with the relationship of each frequency band, it is determined which frequency band the current power density belongs to. When P Band belongs to the low frequency band, the obtained P Band is P LF , which represents the low frequency power and reflects the sympathetic and parasympathetic nerve activity. When P Band belongs to the high frequency band, the obtained P Band is P HF , which represents the high frequency power and reflects the parasympathetic nerve activity. When P Band belongs to the very low frequency band, the obtained P Band is P VLF , which represents the very low frequency power and is related to the long-term regulation mechanism (such as body temperature), and P LF / P HF represents the balance between sympathetic and parasympathetic nerves.

[0295] Specifically, the HRV nonlinear feature extraction can use Poincaré plot analysis, specifically:

[0296]

[0297] where SD1 represents the short-term HRV component, which is related to the parasympathetic nerve activity, and SD2 represents the long-term HRV component, which is related to the sympathetic nerve activity.

[0298] In calculating the heart rate variability related features, the ECG signal in the target vital sign signal is used.

[0299] Three, respiratory rate (RR).

[0300] In calculating the respiratory rate related features, the BIOZ filtered signal in the target vital sign signal is used. The respiratory rate in the BIOZ signal in the target vital sign signal is mainly extracted through the impedance change of the chest, which can be based on time domain and frequency domain methods.

[0301] (a) Time domain method.

[0302] The periodic change of the chest impedance waveform is used to calculate the average respiratory period, and the specific formula is:

[0303]

[0304] where RR represents the respiratory rate, in breaths per minute (bpm), and T breath represents the average period of a single breath, in seconds.

[0305] Feature extraction steps:

[0306] 1. Detect the peak values or zero-crossing points of the respiratory waveform from the BIOZ filtered signal in the target vital sign signal.

[0307] 2. Calculate the time interval T breath between adjacent peak values or zero-crossing points.

[0308] 3. Take the average value of multiple T breath , and substitute it into the formula (RR = 60 / T breath ) for calculation.

[0309] (b) Frequency domain method.

[0310] Based on the fast Fourier transform (FFT) analysis of the frequency spectrum of the BIOZ filtered signal in the target vital sign signal, the respiratory frequency is extracted, and the specific formula is:

[0311] RR = f peak · 60;

[0312] where f peak represents the main peak frequency of the respiratory frequency in the BIOZ signal in the target vital sign signal, in Hz.

[0313] Feature extraction steps:

[0314] 1. Perform FFT on the BIOZ signal in the target vital sign signal.

[0315] 2. Find the main peak f peak in the low frequency band (usually 0.1 Hz-0.5 Hz).

[0316] 3. Substitute into the formula to calculate the respiratory rate RR = f peak · 60.

[0317] Four, blood oxygen (SpO2).

[0318] When calculating the blood oxygen (SpO2) related features, the photoelectric plethysmogram (PPG) filtered signal in the target vital sign signal is used, and the blood oxygen saturation (SpO2) feature extraction algorithm is usually based on the ratio of red light (R) and infrared light (IR) signals, including the following steps:

[0319] a) Signal decomposition:

[0320] Separate the alternating current (AC) and direct current (DC) components from the PPG signal in the target vital sign signal.

[0321] Usually, the AC component is extracted using a high-pass filter, and the DC component is extracted using a low-pass filter.

[0322] b) Calculate the ratio:

[0323] AC / DC is calculated for red and infrared signals, respectively.

[0324] c) Calculate Rratio:

[0325] The formula for calculating Rratio:

[0326]

[0327] Where AC R represents the alternating current (AC) component of the red light signal (reflecting the dynamic part of the pulse wave). DC R represents the direct current (DC) component of the red light signal (reflecting the average or base part of the signal). AC IR represents the alternating current (AC) component of the infrared signal. DCIR represents the direct current (DC) component of the infrared signal.

[0328] d) Calculate SpO2.

[0329] SpO2 calculation formula:

[0330] SpO2 = 100 - R ratio ;

[0331] Where: C represents a proportionality constant, related to the calibration curve, usually determined based on experimental data. This value is determined by experimental methods, usually relying on the true SpO2 value measured by a standard oxygen analyzer, and fitting with the measured value of the device. This value may vary due to factors such as population, skin color, environmental light interference, etc., and needs to be considered in multi-dimensional calibration in practical application.

[0332] Five, body temperature (T).

[0333] In wearable devices, the algorithm of high-precision ultra-low-power digital temperature sensor is usually based on the digital output value of the temperature sensor, and the temperature value T is obtained by calibration and formula calculation.

[0334] For digital temperature sensors, the body temperature calculation formula is usually:

[0335] T = a·Output_Code + b;

[0336] Or (when non-linear calibration):

[0337]

[0338] Where T represents the body temperature value, Output Code represents the digital output value of the temperature sensor, usually the value after ADC conversion or the direct digital register value; a and b represent the calibration coefficients provided by the sensor manufacturer or obtained through calibration test, C i represents the calibration coefficient of polynomial fitting, used for higher precision nonlinear calibration, and n represents the order of the polynomial, usually 2 or 3.

[0339] Six, blood pressure.

[0340] Blood pressure includes: systolic blood pressure (SBP) and diastolic blood pressure (DBP), wherein the systolic blood pressure represents the blood pressure value when the heart contracts, and the diastolic blood pressure represents the blood pressure value when the heart relaxes.

[0341] 1) Pulse wave transit time (PTT) extraction.

[0342] Pulse wave transit time (PTT) refers to the time for the pulse wave emitted from the heart to propagate to the peripheral blood vessels, which has a certain correlation with blood pressure. PTT is calculated based on the time difference between ECG and PPG signals in the target vital sign signal, and is commonly used for blood pressure estimation. PTT extraction is as follows:

[0343] PTT = t PPG_peak -t R ;

[0344] Where t R represents the R wave peak time in the ECG signal in the target vital sign signal, representing the time when the R wave occurs in the electrocardiogram; t PPG_peak represents the pulse wave peak time in the PPG signal in the target vital sign signal, representing the time when the maximum amplitude of the pulse wave occurs; PTT represents the time difference from the R wave peak of the ECG signal to the PPG signal peak.

[0345] 2) Blood pressure estimation.

[0346] Regression model can be used for blood pressure estimation, specifically:

[0347] SBP estimation formula:

[0348] SBP = a1·PTT + a2·HR + a3·PPG features + b1;

[0349] DBP estimation formula:

[0350] DBP = a4·PTT + a5·HR + a6·PPG features + b2;

[0351] Where a1, a 2， a 3， a 4，a 5， a6represents the coefficient of the regression model, representing the contribution of different input features to blood pressure, obtained through training data; b1,b2represents the constant bias term, the constant value in the regression model, PPG features represent other features in the PPG signal, such as the rise time of the pulse wave, the peak amplitude, the fall time, etc.; HR represents the heart rate, representing the number of heartbeats per unit time, usually in BPM (beats per minute).

[0352] After obtaining the estimated blood pressure value, the estimated blood pressure value can also be smoothed to reduce the noise in the estimated blood pressure value. It replaces each data point with the average value of the data in its neighborhood by performing a sliding window average operation on the signal, thereby achieving smoothing of the estimated blood pressure value. The smoothing algorithm can include: Moving Average Filter, Exponential Moving Average (EMA), and Kalman Filter.

[0353] a) Moving Average Filter.

[0354] Moving Average Filter is based on the principle of mean filtering, which calculates the average of consecutive sampling data. It sums and averages the sample values of the input signal within a sliding time window. As the window moves along the time axis, it continuously calculates new average values to obtain the filtered output signal. This can effectively suppress high-frequency noise in the signal and make the signal smoother. The specific formula is:

[0355]

[0356] where, represents the smoothed i-th systolic blood pressure value, N represents the window size, SBP j represents the j-th estimated systolic blood pressure. represents the smoothed i-th diastolic blood pressure, DBP j represents the j-th estimated diastolic blood pressure.

[0357] b) Exponential Moving Average (Exponential Moving Average, abbreviated as: EMA).

[0358] Exponential Moving Average assigns exponentially decreasing weights to past observations, with more recent data having higher weights and more distant data having lower weights. This allows it to respond more quickly to new data and more timely track the trend of data changes. The specific formula is:

[0359]

[0360] where α represents a smoothing factor, typically between 0 and 1, controlling the weight of smoothing.

[0361] c) Kalman Filter.

[0362] Kalman Filter is a recursive filter commonly used to estimate the state of a linear dynamic system from noisy and incomplete data. It is an optimal estimation method that optimizes the estimate by combining the system model and the measurements, including:

[0363] Prediction: Predict the state at the next time step based on the current state estimate and the known system model.

[0364] Update: Use the new measurements to correct the predicted value, resulting in a more accurate estimate.

[0365] The formula of Kalman Filter is:

[0366]

[0367] where, represents represents or z k represents the current measurement SBP i or DBP i ; K k represents the Kalman gain, used to weigh and SBP i accuracy, or, to weigh and DBP i accuracy, A represents the state transition matrix, describing the change of the system, B represents the control matrix, controlling the input to the state, H is the measurement matrix, representing the relationship between the state and the estimated blood pressure value, u k is the control input, used to describe the external influence on the system.

[0368] Seven, arrhythmia (Rhythm).

[0369] When extracting arrhythmia features, the ECG signal in the target vital sign signal is used.

[0370] The arrhythmia feature extraction algorithm mainly includes QRS detection, HR and HRV extraction, and feature analysis for different types of arrhythmia. The goal is to identify and analyze abnormal rhythms or heartbeat patterns in ECG signals. Using these features, classification and detection of normal rhythm and arrhythmia can be achieved through machine learning and other methods. Arrhythmia detection can be targeted and independent monitoring, so it includes QRS monitoring, HR rhythm extraction, HRV rhythm variability extraction, rhythm monitoring feature extraction and classification.

[0371] 1) QRS detection.

[0372] QRS complexes are used to identify the starting point of a heartbeat. Accurate detection of QRS waves is the first step in arrhythmia analysis, as follows:

[0373]

[0374] where ECG[t] represents the ECG signal in the target vital sign signal, represents the derivative of ECG[t], highlighting the sharp changes of QRS waves, represents the integral of , which helps to identify the width and position of QRS waves, thresholding is the set threshold, and QRS(t) represents the position of QRS complexes. This can be achieved through the following steps:

[0375] 1. Signal filtering: First, perform digital filtering on the ECG signal in the target vital sign signal to remove high-frequency noise and baseline drift.

[0376] 2. Derivative and integral: Calculate the derivative and integral of the ECG signal in the target vital sign signal to highlight the features of QRS waves.

[0377] 3. QRS detection: Detect QRS complexes by setting a threshold.

[0378] 2) Heart rate (HR) extraction.

[0379] Heart rate refers to the number of heartbeats per unit time, usually determined by calculating the rr interval (the time difference between two adjacent R wave peaks).

[0380] HR calculation can be achieved through the following steps:

[0381] 1. rr interval calculation:

[0382] R-R interval = t R(i+1) -t R(i) ;

[0383] where t R(i)R-R interval R-R

[0384] 2. Heart Rate (HR) calculation:

[0385] Heart rate is the number of heartbeats per unit time, usually expressed in BPM, and is calculated as:

[0386]

[0387] 3) Heart Rate Variability (HRV) extraction, see above for details in "II. Heart Rate Variability (HRV) features".

[0388] 4) Arrhythmia detection feature extraction and classification.

[0389] For arrhythmia detection, abnormal features (such as PVC, AF, Ventricular Tachycardia, etc.) that are different from normal heart rhythm need to be extracted.

[0390] a) Arrhythmia features.

[0391] • Premature Contractions: refers to QRS waves that are earlier than expected, which can be PVC (Premature Ventricular Contractions) or PAC (Premature Atrial Contractions).

[0392] • Atrial Fibrillation (AF): Atrial fibrillation is usually manifested as irregular QRS waves and ECGs without P waves.

[0393] • Ventricular Tachycardia (VT): usually manifested as a group of QRS waves that are wide and uniform in shape.

[0394] b) Feature extraction:

[0395] 1. R wave morphology analysis:

[0396] Compare the shape of the R wave with the typical normal R wave shape to detect abnormalities.

[0397] Rwavemorphology: identify the presence of premature beats, etc. by comparing the amplitude, width and shape of the R wave.

[0398] 2. QRS complexity analysis:

[0399] Calculate the width and amplitude of the QRS complex.

[0400] RScomplexity: determine the presence of ventricular tachycardia and other abnormalities by analyzing the complexity of the QRS wave.

[0401] 5) Classification algorithm (e.g., SVM or neural network):

[0402] By machine learning methods (such as support vector machine SVM or CNN neural network), combined with extracted features (HR, HRV, R wave form, etc.), normal rhythm and various arrhythmias can be classified. The formula is as follows:

[0403]

[0404] where, represents the predicted rhythm category (normal, atrial fibrillation, premature beat, etc.), f ML represents the machine learning model trained, which is learned by the process of labeling data to classify heart rhythm.

[0405] The process of comparing vital sign parameters with vital sign standard parameters is described in detail below.

[0406] In the VSM (Vital Signs Monitoring) monitoring system, HR (heart rate), RR (respiratory rate), SpO2 (blood oxygen saturation), BP (blood pressure), HRV (heart rate variability) and temperature are important physiological signals that can reflect the health status of individuals. Data analysis and comparison can be achieved through various methods. Next, we will discuss how to perform these data analysis and comparison between different indicators.

[0407] 1. Heart rate (HR) analysis.

[0408] Heart rate is a core indicator for assessing the health status of the heart. It is usually analyzed through the following analysis methods:

[0409] Basic statistical analysis: including mean, standard deviation, maximum and minimum of heart rate, etc.

[0410] Time domain analysis: based on RR interval (heart beat interval), calculate HRV, etc.

[0411] Frequency domain analysis: assess the changes in heart rate through spectral analysis.

[0412] 2. Respiratory rate (RR) analysis.

[0413] Respiratory rate reflects the health status of the respiratory system. Data analysis of RR includes:

[0414] Time domain analysis: by monitoring the number of breaths per minute, calculate its mean, maximum and minimum.

[0415] Frequency domain analysis: use spectral analysis to evaluate the respiratory frequency components, analyze the proportion of low and high frequency components.

[0416] 3. Oxygen saturation (SpO2) analysis.

[0417] SpO2 is a key indicator of oxygenation status, and the analysis method is:

[0418] Time domain analysis: Calculate the mean value of the fluctuation of blood oxygen values.

[0419] Threshold judgment: According to clinical standards, set the alarm threshold below 90% or 92%.

[0420] 4. Blood pressure (BP) analysis.

[0421] Blood pressure (systolic blood pressure SBP and diastolic blood pressure DBP) reflects the health of the heart and blood vessel system. Blood pressure analysis includes:

[0422] Time domain analysis: Calculate the mean, maximum and minimum values of SBP and DBP.

[0423] Fluctuation analysis: Analyze the fluctuation of blood pressure to assess cardiovascular health.

[0424] 5. Heart rate variability (HRV) analysis.

[0425] HRV is a key indicator of autonomic nervous system health, and is usually used:

[0426] Time domain analysis: For example, SDNN (standard deviation) and RMSSD (root mean square of adjacent RR interval differences).

[0427] Frequency domain analysis: By analyzing the ratio of low frequency (LF) and high frequency (HF), the sympathetic and parasympathetic nerve activity is evaluated.

[0428] 6. Temperature (T) analysis.

[0429] Body temperature is an important physiological indicator of body condition, reflecting the health of the immune system and metabolic activity. Body temperature analysis method:

[0430] Time domain analysis: Calculate the mean value and fluctuation of body temperature.

[0431] Change rate: Monitor the change of body temperature over time and find the sudden change of body temperature.

[0432] Data comparison and comprehensive analysis.

[0433] 1) Data correlation analysis.

[0434] By Pearson correlation coefficient or Spearman correlation coefficient, etc., analyze the correlation between HR, RR, SpO2, BP, HRV and T data. For example:

[0435] HR and HRV: High heart rate usually accompanied by low HRV, indicating strong sympathetic nerve activity.

[0436] HR with BP: High heart rate can be accompanied by elevated blood pressure, especially systolic pressure.

[0437] HR with SpO2: When oxygen saturation decreases, heart rate usually increases to compensate for the lack of oxygen supply.

[0438] T with HR / SpO2: Changes in body temperature can affect heart rate and oxygen saturation, especially in fever or cold environments.

[0439] 2) Multidimensional data analysis and trend comparison.

[0440] Long-term trend analysis: Analyze the long-term trends of HR, RR, SpO2, BP, HRV and T, for example, a gradual increase in heart rate and a gradual decrease in oxygen saturation may indicate certain health problems.

[0441] Short-term fluctuation analysis: By monitoring short-term fluctuations, possible sudden health risks can be found, such as a sudden increase in body temperature may indicate fever or infection.

[0442] 3) Risk prediction and health assessment.

[0443] Risk assessment model: Based on multiple physiological indicators (such as HR, RR, SpO2, BP, HRV and T), a machine learning model is built to predict cardiovascular events, respiratory failure or infection risk. For example, high HR, low HRV, low SpO2 and high T may indicate potential infection.

[0444] Comprehensive health index: Calculate a comprehensive health index using all physiological indicators to detect health problems early.

[0445] Integrating HR, RR, SpO2, BP, HRV and T data can comprehensively assess an individual's physiological state and health risks. Through correlation analysis, trend analysis and multidimensional data comparison, potential health problems can be more accurately identified and warned.

[0446] The following details the analysis and comparison of arrhythmia Rhythm data.

[0447] 1) Arrhythmia analysis.

[0448] Normal heart rhythm (normal sinus rhythm): In normal circumstances, heart rate should be 60-100 BPM, and QRS complex is regular.

[0449] Atrial fibrillation (AF): Irregular heart rate, common range is 100-175 BPM. At this time, the electrical activity of the atrium loses regularity, and the ventricular beat is irregular.

[0450] Premature Ventricular Contractions (PVCs): PVCs usually appear as early abnormal QRS complexes, leading to irregular rr intervals, but generally do not affect the overall heart rate.

[0451] Ventricular Tachycardia (VT): VT is characterized by rapid ventricular contractions, resulting in a significantly increased heart rate, typically between 100-250 BPM.

[0452] 2) Arrhythmia Classification.

[0453] Arrhythmia classification can be based on extracted features. Common classification methods include: Support Vector Machines (SVM), Neural Networks, Decision Trees, etc. Feature extraction is as follows:

[0454] Premature Contractions:

[0455] PVCs (Premature Ventricular Contractions) or PACs (Premature Atrial Contractions).

[0456] Features: Abnormal QRS complexes, narrower or wider, with different shapes.

[0457] Atrial Fibrillation (AF):

[0458] Features: Irregular R-wave complexes, no obvious P-wave.

[0459] During AF, rr intervals are irregular, with large fluctuations.

[0460] Ventricular Tachycardia (VT):

[0461] Features: Wide and uniform QRS complexes, faster heart rate.

[0462] 3) Machine Learning Classification.

[0463] Machine Learning Classification (Heart Rhythm Classification)

[0464] Machine learning classification is used to classify heart rhythms based on features extracted from ECG signals. Common machine learning algorithms include Support Vector Machines (SVM), Random Forests (RF), Neural Networks, etc.

[0465] ① Feature Input:

[0466] Features extracted from the ECG signal of the target vital sign signal are used to input into the machine learning model for classification. Common features include:

[0467] Heart rate (HR), heart rate variability (HRV), QRS complex features (such as amplitude, width, etc.), P-wave and T-wave features, rr intervals.

[0468] ② Machine learning model:

[0469] Training phase:

[0470] X train : Training dataset containing features and corresponding labels (heart rhythm types).

[0471] Y train : Labels in the training dataset (such as normal, atrial fibrillation, ventricular premature beat, etc.).

[0472] f ML (Xtrain): Trained machine learning model.

[0473] Classification phase:

[0474] Xnew: New input data (extracted features).

[0475] Predicted heart rhythm type, for example: normal rhythm, atrial fibrillation, ventricular premature beat, etc., model output.

[0476] Model formula:

[0477]

[0478] Comparison of arrhythmias.

[0479] Comparison of arrhythmias mainly distinguishes different arrhythmia types by comparing the differences between features, common methods include comparing heart rate (HR), rr interval, etc. Common arrhythmia feature comparison:

[0480] ① Normal rhythm:

[0481] · Normal ECG signal presents regular rr interval, stable heart rate.

[0482] · HRnormal: Normal heart rate, usually 60-100 BPM.

[0483] · Atrial fibrillation (AF):

[0484] · Features: irregular rr interval, lack of P wave, unstable heart rate.

[0485] · Atrial fibrillation has large fluctuations in rr interval, frequent changes.

[0486] · Ventricular premature beat (PVC):

[0487] · Features: QRS complex appears in advance, abnormal shape.

[0488] • Irregular RR intervals with PVCs

[0489] • Ventricular Tachycardia (VT):

[0490] • Characteristics: Wide QRS complex, very fast heart rate

[0491] • HRVT: High rate VT, typically > 150 BPM

[0492] ② Comparison Analysis Formula:

[0493] RR Interval Difference:

[0494] To compare RR intervals under different heart rhythms and determine the type of arrhythmia, the formula is as follows:

[0495] ΔR-R = |(R-R current ) - (R-R baseline ) |;

[0496] Where ΔR-R represents the difference between the current RR interval and the baseline interval, used to identify heart rate abnormalities; R-R current represents the current detected RR interval, and R-R baseline represents the baseline normal heart rhythm RR interval.

[0497] Heart Rate Difference (HR):

[0498] To compare the changes in heart rate, especially the fluctuations during abnormal heart rhythms, the formula is:

[0499] ΔHR = |HR current - HR baseline |;

[0500] Where ΔHR represents the difference between the current heart rate and the baseline heart rate, used to identify heart rhythm abnormalities; HR current represents the current measured heart rate; and HR baseline represents the normal heart rate (usually 60-100 BPM).

[0501] Ratio Feature: Use the ratio of HR and RR interval to further compare different types of arrhythmias.

[0502]

[0503] Where Ratio represents the ratio feature used to distinguish between arrhythmia types.

[0504] In arrhythmia analysis, machine learning classification methods predict rhythm types by extracting features of ECG signals such as HR, HRV, QRS features, etc. In arrhythmia comparison analysis, by comparing features such as r-r interval, heart rate, etc. under different rhythm states, different types of arrhythmia (such as normal, atrial fibrillation, premature ventricular contraction, ventricular tachycardia, etc.) can be further distinguished.

[0505] The arrhythmia Rhythm standard is shown in Table 6:

[0506] Table 6

[0507]

[0508] It is worth noting that the hardware design in this disclosure conforms to YY9706.102-2021 "Medical electrical equipment, Part 1-2: General requirements for basic safety and basic performance parallel standards: electromagnetic compatibility requirements and tests"; conforms to GB 9706.1-2020 "Medical electrical equipment, Part 1: General requirements for basic safety and basic performance"; conforms to GB 9706.227-2021 "Medical electrical equipment, Part 2-27: Special requirements for applicable requirements for basic safety and basic performance of electrocardio monitoring equipment"; the hardware design conforms to YY9706.247-2021 "Medical electrical equipment, Part 2-47: Special requirements for applicable requirements for basic safety and basic performance of dynamic electrocardiogram system".

[0509] Key technical advantages in this disclosure:

[0510] 1. Non-networked design: real-time data analysis, alarm, privacy and local storage and monitoring report, completely independent without network operation, ensuring data security.

[0511] 2. Vital signs parameters (such as: heart rate (HR), respiratory rate (RR), oxygen saturation (SpO2), T (body temperature), HRV (heart rate variability), Rhythm arrhythmia (AF (atrial fibrillation), PVC (premature ventricular contraction), VT (wet tachycardia) and other monitoring and early warning prompts to the wearer or carrier themselves.

[0512] 3. According to age, gender, height, weight, altitude (3000m; or choose >3000m).

[0513] 4. Monitoring type: automatic monitoring or targeted monitoring (such as arrhythmia, myocardial infarction, etc.).

[0514] 5. Portability: small volume, light weight, suitable for carrying, suitable for various environments.

[0515] 6. High energy efficiency: low power design and high efficiency power management, supporting long continuous operation.

[0516] Advantages of the present disclosure

[0517] 1. Users can grasp the health status in time, effectively prevent acute diseases, and have a wide market application prospect. The system does not depend on external network or external system, and has the ability of independent operation. In the environment without network, the environment with invalid network scheme, and the environment requiring privacy protection, the application of the intended purpose is ensured.

[0518] 2. The embedded algorithm can quickly and effectively diagnose vital signs, reducing the problems of misdiagnosis and missed diagnosis caused by delay.

[0519] 3. Real-time: when the user's heart rate, blood oxygen saturation, and respiratory rate value are abnormal, timely visual, auditory, and tactile awakening is realized, and lifesaving time is gained at critical moment.

[0520] 4. Low power design is adopted, long time endurance is achieved, and the device is suitable for long time wearing scene.

[0521] 5. The device also has human-computer interaction function, which is convenient for human-computer voice interaction setting.

[0522] 6. The device has SOS function which can be selected by the user.

[0523] 7. The device is suitable for wide application in user groups such as chronic disease patients, the elderly, sports enthusiasts, and is suitable for various scenes such as family, hospital, work and outdoor sports.

[0524] 1) Emergency rescue: vital sign monitoring in disaster area, wild or battlefield.

[0525] 2) Hospital monitoring: as a supplementary device for hospital patient vital sign monitoring.

[0526] 3) Home care: suitable for health monitoring of chronic disease patients or the elderly.

[0527] 4) Remote area: no need to rely on network, can be independently operated in remote area.

[0528] 5) High altitude area: for example, above 3000m.

[0529] 6) Protection population: personal vital sign value for national dignitaries and people who need privacy protection, no network leakage, monitoring and early warning only for themselves.

[0530] In one embodiment, as shown in Figure 3 the medical grade VSM dynamic monitoring system in the present disclosure further comprises a warning prompt module 7;

[0531] The early warning prompt module 7 is connected with the data processing module 3 and the power management module 5 respectively;

[0532] The data processing module 3 is further configured to generate an early warning signal when an abnormal vital sign parameter is determined according to the comparison result, and send the early warning signal to the early warning prompt module 7.

[0533] The early warning prompt module 7 outputs the early warning signal.

[0534] In one embodiment, the early warning prompt module 7 can have the function of voice broadcast prompt of vital sign measurement value, for example: visual, auditory or tactile early warning prompt.

[0535] At this time, in terms of outputting the early warning signal, the early warning prompt module 7 is specifically configured to:

[0536] Output the early warning signal in the form of vision, for example: output the early warning signal through a three-color LED or a display screen, etc., and when the early warning signal is output through the three-color LED, different early warning levels can be set according to different light colors:

[0537] Green (normal): all parameters are within the normal range.

[0538] Yellow (warning): some parameters are close to the early warning threshold, but have not exceeded the threshold (such as HR slightly higher than the normal range, but has not reached the dangerous level).

[0539] Red (emergency): some parameters exceed the early warning threshold, which needs to be paid attention to immediately (such as SpO2 decreases to below 85%, and HR is lower than 50 BPM).

[0540] Or,

[0541] Output the early warning signal in the form of hearing, for example: output the early warning signal through voice (including but not limited to Chinese, English) or buzzer, loudspeaker, etc.

[0542] Or,

[0543] Output the early warning signal in the form of touch, for example: output the early warning signal through a vibration motor.

[0544] In the VSM (Vital Signs Monitoring) system, after data preprocessing and feature extraction, data analysis and comparison are completed, the generation of early warning system and monitoring report is the next key step. For each monitoring parameter (such as HR, RR, SpO2, BP, HRV, T, AF, PVC, VT), the early warning and monitoring report can be designed through the following steps.

[0545] The vital signs VSM standard parameter reference table is shown in Tables 2 and 3. The normal physiological value range of the vital signs VSM standard parameters is defined as the early warning threshold value for the age group, static state, static rest state, gait state, running state, falling state, and rapid motion state.

[0546] The following methods can be used to trigger early warnings:

[0547] Rule-based early warning: based on pre-defined threshold rules, an alarm is triggered when a parameter exceeds the threshold value.

[0548] For example: if HR > 100 BPM, a high heart rate alarm is triggered.

[0549] Machine learning early warning: by training a machine learning model, patterns in different states are learned from historical data to identify potential dangerous signals. For example, using a support vector machine (SVM) or a random forest (RandomForest) to comprehensively analyze HR, SpO2, and BP data to determine whether there is an abnormal pattern.

[0550] In the present disclosure, early warnings can be made in a timely manner through the early warning prompt module 7, so that the user can perform corresponding emergency strategies based on the early warning signal, which will put the early warning or emergency measures in the hands of the wearer himself, avoiding missed treatment.

[0551] Through the above analysis, it can be seen that the present disclosure provides a completely local (without network) solution, which uses biometric (such as photoplethysmography PPG) and bioimpedance BIOZ, electrocardiogram ECG) sensors, digital temperature sensors to collect data, and a data processing module 3 to independently analyze real-time data, complete real-time collection, processing, and early warning or monitoring reporting to the wearer, taking into account the portability and high efficiency, low energy consumption design of wearable, improving the real-time, accuracy of monitoring, and ensuring data privacy, intervening in advance before an unexpected health event occurs, meeting the data security and high reliability requirements in different application scenarios.

[0552] Core elements of the present disclosure:

[0553] 1. The solution focuses on completely local (without external network) real-time vital sign data analysis, alarm, privacy, and local storage and monitoring reporting.

[0554] 2. Vital signs (VSM) monitoring: including but not limited to: heart rate (HR), respiratory rate (RR), blood oxygen saturation (SpO2), T (body temperature) HRV (heart rate variability), Rhythm arrhythmia (AF (atrial fibrillation), PVC (ventricular premature beat), VT (wet tachycardia)).

[0555] 3. Targeted monitoring and early warning, for patients diagnosed with rhythm disorders, heart attack, etc. Implement targeted monitoring and early warning.

[0556] 4. No need for external analysis graphics, based on internal analysis and model verification technology. Means that non-professionals can also directly read the monitoring report results.

[0557] 5. Collecting and correlating with patient (age, gender, height, weight, altitude, arrhythmia patient, etc. Basic information), exercise state, etc. Increase the quality of analysis and processing.

[0558] 6. Based on the current collection, early warning of vital signs (VSM) risk trend changes.

[0559] 7. Local paperless monitoring report.

[0560] 8. The scheme provides a reliable, private and autonomous continuous (such as 1-15 days) health monitoring method, without external network protection for users privacy, private and autonomous continuous health monitoring method.

[0561] 9. The scheme ensures that no matter what environment the user is in, such as: exercise, indoor, outdoor, work, sleep, nursing home, medical institution, etc., as well as high altitude, war, natural disaster, etc. Environment, accurate health information and instant feedback can be obtained for the wearer himself.

[0562] 10. Hardware combination: vital signs sensor (i.e. wearable sensor 11 in the above embodiments), analog front end 12, and data processing module 3 (may include FPU-MCU or RNN-MCU or CNN-MCU or AI MCU) etc. Build hardware combination structure. At least one or more FPU-MCU (or RNN-MCU or CNN-MCU or AI MCU) centralized processing or separate processing such as heart rate, respiration rate, blood oxygen vital signs data filtering, feature extraction, motion artifact, data judgment, etc. Analysis. The microprocessor involved is ultra-low power, high efficiency, high reliability (complying with medical-grade reliability applications).

[0563] 11. The scheme provides a hardware requirement of ±50mVP-p~±200mVP-p AC dynamic input range to prevent measurement saturation (signal integrity), and realizes acceleration motion artifact processing algorithm.

[0564] 12. Independent use or combined use of acceleration, gyroscope, magnetometer, combined with 5, to realize independent motion artifact processing algorithm-acceleration VSM signal preprocessing algorithm, to realize one of the core algorithms of acceleration participating in filtering.

[0565] 13. The embedded convolutional neural network vital signs (VSM) feature value extraction algorithm is the second core algorithm. However, single or small number of parameters feature extraction is not limited to using other methods.

[0566] 14. Automatically and accurately match the motion state: static state, static rest state, gait state, running state, falling state, fast motion state, without the need for the wearer to set.

[0567] 15. Vital signs (VSM) monitoring scheme can manufacture single parameter monitoring devices (such as: respiratory rate (RR)), and can also manufacture vital signs (VSM) multi-parameter form monitoring devices (such as: including but not limited to: heart rate (HR), respiratory rate (RR), blood oxygen saturation (SpO2), T (body temperature), HRV (heart rate variability), rhythm arrhythmia (AF (atrial fibrillation), PVC (ventricular premature beat), VT (wet tachycardia)).

[0568] 16. Extensible heart sound, lung breath sound, etc.

[0569] 17. Extensible to manufacture local non-wearable vital signs (VSM) early warning monitoring devices, making physical examination comprehensive indicators.

[0570] ​​​​​Based on the same inventive concept, the embodiments of the present application also provide a medical-grade VSM dynamic monitoring method for the medical-grade VSM dynamic monitoring system as mentioned above. The implementation scheme for solving the problem provided by the method is similar to the implementation scheme described in the system above, so the specific limitations in one or more medical-grade VSM dynamic monitoring method embodiments provided below can refer to the limitations of the medical-grade VSM dynamic monitoring system described above, which will not be repeated here. In an exemplary embodiment, the present disclosure also provides a medical-grade VSM dynamic monitoring method, which is applied to a medical-grade VSM dynamic monitoring system, the medical-grade VSM dynamic monitoring system comprising: a sensing module, an inertial measurement module, a data processing module, a display and interaction module, a pre-warning prompt module and a power management module, the data processing module being connected with the inertial measurement module and the display and interaction module respectively; the power management module being connected with the inertial measurement module, the data processing module and the display and interaction module respectively; the method comprising: the display and interaction module acquires the basic information of the user to be monitored, the basic information comprising: age, gender, height, weight, altitude and monitoring mode; the monitoring mode comprising: automatic monitoring or selective targeted monitoring; the sensing module measures the vital sign signal and sends the vital sign signal to the data processing module; the inertial measurement module measures the inertial signal and sends the inertial signal to the data processing module; the data processing module acquires the vital sign parameter of the user to be monitored according to the inertial signal and the vital sign signal, acquires the vital sign standard parameter according to the basic information and the inertial signal, compares the vital sign parameter with the vital sign standard parameter, generates a monitoring report according to the comparison result, and sends the monitoring report to the display and interaction module; the display and interaction module outputs the monitoring report.

[0571] In one embodiment, the medical-grade VSM dynamic monitoring system further comprises a pre-warning prompt module; the pre-warning prompt module is connected with the data processing module and the power management module respectively; the method further comprises: when the data processing module determines that an abnormal vital sign parameter occurs according to the comparison result, generating a pre-warning signal and sending the pre-warning signal to the pre-warning prompt module; the pre-warning prompt module outputs the pre-warning signal. In one embodiment, the medical-grade VSM dynamic monitoring system further comprises an SOS rescue module; the SOS rescue module is connected with the data processing module and the power management module respectively; the method further comprises: when the data processing module determines that an abnormal vital sign parameter occurs according to the comparison result, generating a help-seeking signal and sending the help-seeking signal to the SOS rescue module; after receiving the help-seeking signal, the SOS rescue module executes a preset help-seeking means, and the preset help-seeking means includes: sending a distress message to a preset emergency contact person, and / or sending a distress message to a rescue agency.

[0572] In one embodiment, the sensing module comprises a wearable sensor and an analog front end; the analog front end is connected with the wearable sensor, the data processing module and the power management module respectively; the method further comprises: the wearable sensor collects raw vital sign signals and sends the collected raw vital sign signals to the analog front end; the analog front end pre-processes the received raw vital sign signals to obtain the vital sign signals.

[0573] In one embodiment, the vital sign signals include at least one of the following signals: ECG signals, PPG signals, BIOZ signals and temperature signals. In one embodiment, the vital sign parameters include at least one of the following parameters: heart rate, heart rate variability, respiratory rate, blood oxygen saturation, body temperature, blood pressure and arrhythmia. In one embodiment, the output of the pre-warning signal includes: outputting the pre-warning signal in a visual form; or outputting the pre-warning signal in an auditory form; or outputting the pre-warning signal in a tactile form.

[0574] In one embodiment, the method further comprises: the data processing module acquires a medical advice according to the abnormal vital sign parameter, and the medical advice provides measures for the user to be monitored to cope with the abnormal vital sign parameter; the monitoring report includes the vital sign parameters, the abnormal vital sign parameters and the medical advice.

[0575] In one embodiment, the method further comprises: the data processing module acquires a body mass index of the user to be monitored according to the height and the weight; the data processing module acquires a body mass index coefficient and an altitude coefficient according to a target basic parameter; the target basic parameter comprises: the body mass index, or the altitude; the data processing module optimizes the inertial signal according to the body mass index coefficient and the altitude coefficient; and the data processing module determines the vital sign parameter of the user to be monitored according to the optimized inertial signal and the vital sign signal.

[0576] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as there is no contradiction, all combinations shall be considered within the scope of the present disclosure. The principles and implementation manners of the present application are described by using specific examples, and the above embodiments are only used to help understand the method and core idea of the present application; meanwhile, for those skilled in the art, the specific implementation manners and application ranges can be changed according to the idea of the present application. In conclusion, the content of the present disclosure should not be understood as a limitation of the present application.

Claims

1. A medical grade VSM dynamic monitoring system, characterized in that, The medical-grade VSM dynamic monitoring system comprises a sensing module, an inertial measurement module, a data processing module, a display and interaction module, and a power management module; The data processing module is connected with the sensing module, the inertial measurement module, and the display and interaction module respectively; and the power management module is connected with the inertial measurement module, the data processing module, and the display and interaction module respectively; The display and interaction module is configured to acquire basic information of a user to be monitored, wherein the basic information comprises age, gender, height, weight, altitude, and a monitoring mode; and the monitoring mode comprises automatic monitoring or selective targeted monitoring. The sensing module is configured to measure vital sign signals of the user to be monitored, and send the vital sign signals to the data processing module. The inertial measurement module is configured to measure inertial signals of the user to be monitored, and send the inertial signals to the data processing module. The data processing module is configured to acquire vital sign parameters of the user to be monitored according to the inertial signals and the vital sign signals, acquire standard vital sign parameters according to the basic information and the inertial signals, compare the vital sign parameters with the standard vital sign parameters, generate a monitoring report according to a comparison result, and send the monitoring report to the display and interaction module. The display and interaction module is configured to output the monitoring report. The data processing module is specifically configured to: acquire a body mass index of the user to be monitored according to the height and the weight; acquire a body mass index coefficient and an altitude coefficient according to a target basic parameter; the target basic parameter comprises the body mass index or the altitude; optimize the inertial signals according to the body mass index coefficient and the altitude coefficient; and determine the vital sign parameters of the user to be monitored according to the optimized inertial signals and the vital sign signals.

2. The VSM dynamic monitoring system of claim 1, wherein, Further comprising: a warning prompt module; The warning prompt module is connected with the data processing module and the power management module respectively. The data processing module is further configured to generate a warning signal when an abnormal vital sign parameter is determined according to the comparison result, and send the warning signal to the warning prompt module. The warning prompt module is configured to output the warning signal.

3. The VSM dynamic monitoring system of claim 2, wherein, Further comprising: an SOS rescue module; The SOS rescue module is connected with the data processing module and the power management module respectively. The data processing module is further configured to generate a help signal when an abnormal vital sign parameter is determined according to the comparison result, and send the help signal to the SOS rescue module. The SOS rescue module is configured to execute a preset help means after receiving the help signal, wherein the preset help means comprises sending a help-seeking information to a preset emergency contact person and / or sending a help-seeking information to a rescue agency.

4. The VSM dynamic monitoring system of claim 1, wherein, The sensing module comprises a wearable sensor and an analog front end. The analog front end is connected with the wearable sensor, the data processing module, and the power management module respectively. The wearable sensor is configured to collect a raw vital sign signal and send the collected raw vital sign signal to the analog front end. The analog front end is configured to pre-process the received raw vital sign signal to obtain the vital sign signal.

5. The VSM dynamic monitoring system of claim 1, wherein, The vital sign signal includes at least one of an ECG signal, a PPG signal, a BIOZ signal, and a temperature signal.

6. The VSM dynamic monitoring system of claim 1, wherein, The vital sign parameter includes at least one of a heart rate, a heart rate variability, a respiratory rate, an oxygen saturation, a body temperature, a blood pressure, and an arrhythmia.

7. The VSM dynamic monitoring system of claim 2, wherein, In the aspect of outputting the early warning signal, the early warning prompt module is specifically configured to: output the early warning signal in a visual form; or, output the early warning signal in an auditory form; or, output the early warning signal in a tactile form.

8. The VSM dynamic monitoring system of claim 2, wherein, The data processing module is further configured to: obtain a medical suggestion according to the abnormal vital sign parameter, the medical suggestion providing a measure for the user to be monitored to cope with the abnormal vital sign parameter; The monitoring report includes: the vital sign parameter, the abnormal vital sign parameter, and the medical suggestion.

9. A medical grade VSM dynamic monitoring method, characterized in that, The method is applied to the medical-grade VSM dynamic monitoring system as claimed in any one of claims 1-8, the medical-grade VSM dynamic monitoring system including a sensing module, an inertial measurement module, a data processing module, a display and interaction module, an early warning prompt module, and a power management module, the data processing module being connected to the inertial measurement module and the display and interaction module, the power management module being connected to the inertial measurement module, the data processing module, and the display and interaction module, the method including: The display and interaction module obtains basic information of a user to be monitored, the basic information including age, gender, height, weight, altitude, and a monitoring mode; the monitoring mode including automatic monitoring or selective targeted monitoring; The sensing module measures a vital sign signal and sends the vital sign signal to the data processing module; The inertial measurement module measures an inertial signal and sends the inertial signal to the data processing module; The data processing module obtains a vital sign parameter of the user to be monitored according to the inertial signal and the vital sign signal, obtains a vital sign standard parameter according to the basic information and the inertial signal, compares the vital sign parameter with the vital sign standard parameter, generates a monitoring report according to the comparison result, and sends the monitoring report to the display and interaction module; The display and interaction module outputs the monitoring report; The data processing module obtains a body mass index of the user to be monitored according to the height and the weight, obtains a body mass index coefficient and an altitude coefficient according to a target basic parameter, the target basic parameter including the body mass index or the altitude, optimizes the inertial signal according to the body mass index coefficient and the altitude coefficient, and determines the vital sign parameter of the user to be monitored according to the optimized inertial signal and the vital sign signal.

Images