Multi-point non-invasive blood pressure monitoring system and method

Abstract

The invention discloses a multi-point non-invasive blood pressure monitoring system and method. The blood pressure monitoring system comprises an intelligent wearing device and two ear hanging type Bluetooth earphones which are in communication connection. The intelligent wearing device is used for being worn on human tissue, a microprocessor and a communication module are carried in the intelligent wearing device, a display screen and at least one first sensor are carried outside the intelligent wearing device, and the intelligent wearing device is used for obtaining a first detection signal and transmitting the first detection signal to the microprocessor. The two ear hanging type Bluetooth earphones are worn on the two ears respectively, and each Bluetooth earphone is internally provided with at least one second sensor so as to obtain second detection signals of the ear areas respectively and transmit the second detection signals to the microprocessor. The microprocessor is internally provided with a preset algorithm so as to calculate a blood pressure prediction result according to the blood pressure characteristics and display the result in real time through the display screen. According to the invention, multi-point measurement is matched with multiple signals, so that the blood monitoring accuracy is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to a multi-point non-invasive blood pressure monitoring system, and also relates to a corresponding blood pressure monitoring method, and belongs to the technical field of medical devices. BACKGROUND

[0002] Currently, non-invasive blood pressure monitoring devices mainly include wrist type, ear type and upper arm type, each type has its own unique working principle. Wrist type blood pressure monitoring devices often use oscillographic method and photoelectric method. Oscillographic method is to block the arterial blood flow by inflating the cuff at the wrist, then slowly deflating, detecting the change of pressure oscillation wave in the cuff, when the oscillation wave reaches the maximum value, the corresponding pressure is the mean arterial pressure, and then calculating the systolic pressure and diastolic pressure through a specific algorithm. The principle of photoelectric method is to use a photoelectric plethysmogram (PPG) sensor to emit light of a specific wavelength, which is absorbed by the blood in the blood vessels, as the volume of blood changes periodically with the heartbeat, the intensity of reflected or transmitted light also changes accordingly, thus obtaining the pulse wave signal, and estimating the blood pressure value according to the relationship between the characteristic parameters of the pulse wave and the blood pressure. Ear type blood pressure monitoring devices usually use the rich blood vessels and shallow characteristics of the ear, and use photoelectric sensors to detect the pulse wave signal of the ear. The blood circulation of the ear is relatively rich, and the skin of the ear is thin, which is convenient for light signal penetration and detection. Its working principle is similar to the photoelectric method of wrist type, by analyzing the characteristics of ear pulse wave such as pulse wave propagation time, amplitude, etc., and combining with the corresponding algorithm to calculate the blood pressure value. The upper arm type blood pressure monitoring device mostly uses oscillographic method as the main measurement principle. When in use, the cuff is tightly wrapped around the upper arm, the cuff is inflated by air pump, the pressure in the cuff is higher than the systolic pressure, so as to block the brachial artery blood flow, then slowly deflating, the pressure in the cuff gradually decreases, when the pressure is slightly lower than the systolic pressure, the blood flow rushes through the compressed blood vessels to form vortex, generating oscillation wave, the oscillation wave is detected by pressure sensor and converted into electrical signal, the signal is analyzed and processed by microprocessor, and the systolic pressure, diastolic pressure and mean arterial pressure are calculated according to the change rule of oscillation wave.

[0003] However, the existing non-invasive blood pressure monitoring devices have the following deficiencies in measurement accuracy, applicable population and use convenience, etc.

[0004] In terms of measurement accuracy, the measurement method with the photoelectric sensor as the core is disturbed by various factors. For example, the photoelectric sensor performs differently in different skin colors of human tissues. For people with darker skin, the melanin in their skin absorbs more light signals, resulting in a decrease in the intensity of reflected light or transmitted light, thereby affecting the accuracy of the pulse wave signal and reducing the accuracy of blood pressure measurement. For people of different ages, the physiological characteristics such as blood vessel elasticity and skin thickness also affect the measurement effect of the photoelectric sensor. The blood vessel elasticity of the elderly decreases, the blood vessel wall thickens, which changes the propagation characteristics of the pulse wave and the absorption and scattering of the light signal; the blood vessels of children are thinner, and the skin is tender, which responds to the light signal differently from adults, which increases the difficulty of accurately measuring blood pressure.

[0005] From the perspective of the applicable population, existing devices are difficult to meet the needs of all populations. For patients with severe arteriosclerosis, arrhythmia and other diseases, the traditional non-invasive blood pressure monitoring method may not accurately reflect their true blood pressure. Arteriosclerosis will make the blood vessel wall hard and the elasticity decrease, resulting in changes in the propagation speed and shape of the pulse wave, and the conventional measurement algorithm based on the characteristics of the pulse wave cannot adapt to such changes, thereby causing measurement errors. The heart rhythm of patients with arrhythmia is irregular, and the period and amplitude of the pulse wave also show irregular changes, making it difficult for the blood pressure measurement method based on periodic pulse wave analysis to accurately calculate the blood pressure value.

[0006] In terms of convenience of use, some devices have certain limitations. For example, although the upper arm type blood pressure monitoring device has relatively high measurement accuracy, it needs to wrap the entire upper arm due to its large volume, which is not very convenient to use in some situations, such as during exercise or when the arm needs to be frequently moved, which will greatly restrict the user and affect normal activities. While the wrist type and ear type devices are relatively portable, their stability and comfort in wearing still need to be improved. The wrist type device is prone to displacement when the wrist moves, which affects the contact between the sensor and the skin and affects signal acquisition. SUMMARY

[0007] The primary technical problem to be solved by the present application is to provide a multi-point non-invasive blood pressure monitoring system.

[0008] Another technical problem to be solved by the present application is to provide a multi-point non-invasive blood pressure monitoring method.

[0009] To achieve the above technical purposes, the present application adopts the following technical solutions:

[0010] According to a first aspect of an embodiment of the present application, a multi-point non-invasive blood pressure monitoring system is provided, comprising a smart wearable device and an ear-hanging Bluetooth earphone, the smart wearable device being in communication connection with the ear-hanging Bluetooth earphone; wherein,

[0011] The intelligent wearable device is worn on a specific tissue of a human body and adheres to the skin of the human body, and at least one first sensor is arranged on the intelligent wearable device to acquire a first detection signal of a region where the specific tissue is located;

[0012] The over-ear Bluetooth earphones are respectively worn on the two ears, and at least one second sensor is arranged on the over-ear Bluetooth earphones to acquire a second detection signal of the two ears;

[0013] The internal part of the intelligent wearable device is further provided with a microprocessor and a communication module, the microprocessor is in communication connection with the first sensor and the second sensor respectively through the communication module to receive the first detection signal and the second detection signal, and multi-site feature fusion is performed to form a blood pressure feature, so that a blood pressure prediction result is calculated based on a preset algorithm according to the blood pressure feature.

[0014] Preferably, the intelligent wearable device is a smart watch, which is used to be worn on a wrist; the at least one first sensor includes a wrist PPG sensor and a wrist ECG sensor;

[0015] The external part of the smart watch is further provided with a display screen, which is connected with the microprocessor and used to display the blood pressure prediction result; the wrist PPG sensor adheres to the skin of the wrist to acquire a wrist PPG signal; the wrist ECG sensor adheres to the skin of the wrist to acquire a wrist ECG signal; wherein the wrist PPG sensor and the wrist ECG sensor are in communication connection with the communication module;

[0016] The at least one second sensor includes an ear PPG sensor and an ear ECG sensor, the ear PPG sensor adheres to the skin of the ear to acquire an ear PPG signal; the ear ECG sensor adheres to the skin of the ear to acquire an ear ECG signal; wherein the ear PPG sensor and the ear ECG sensor are in communication connection with the communication module;

[0017] The microprocessor receives the wrist PPG signal, the wrist ECG signal, the ear PPG signal and the ear ECG signal through the communication module, and after signal preprocessing and feature extraction, multi-site feature fusion is performed to form a blood pressure feature; and the blood pressure feature is taken as an input to calculate a blood pressure prediction result through a preset model and a Kalman weighting algorithm, and the blood pressure prediction result is displayed in real time through the display screen;

[0018] The preset model adopts a neural network, and a large amount of historical blood pressure data and corresponding physiological signal data are used for model training to construct an LSTM model for predicting blood pressure; the LSTM model is used to output a blood pressure prediction result according to the wrist PPG signal, the ear PPG signal, the wrist ECG signal and the ear ECG signal.

[0019] The Kalman weighting algorithm is embedded in the preset model, and is used to real-time assign different weights to the wrist PPG signal and the ear PPG signal according to a preset logic, so as to real-time correct the blood pressure prediction result.

[0020] Preferably, the wrist PPG sensor at least includes a pressure diaphragm sensor and / or a PMUT sensor.

[0021] The pressure diaphragm sensor is used to obtain the micro pressure change of the wrist skin surface caused by pulse jumping, and convert the pressure change into a first electrical signal, so as to obtain the wrist PPG signal based on the first electrical signal.

[0022] The PMUT sensor is used to obtain the ultrasonic wave change of the wrist skin surface caused by pulse jumping, and convert the ultrasonic wave change into a second electrical signal, so as to obtain the wrist PPG signal based on the second electrical signal.

[0023] The wrist ECG sensor detects the electro-physiological signal of the wrist by contacting the wrist skin, so as to obtain the wrist ECG signal based on the electro-physiological signal of the wrist.

[0024] Preferably, the first sensor includes:

[0025] An ultrasonic detection part is used to contact the human skin of a specific tissue, and is used to emit or receive ultrasonic waves, so as to obtain the blood vessel diameter change.

[0026] A pulse wave detection part is arranged on the side of the ultrasonic detection part away from the human skin, and is used to detect the pulse wave signal.

[0027] At the same time, the blood vessel diameter change matches the pulse wave signal, so as to correct the pulse wave signal based on the blood vessel diameter change.

[0028] Preferably, the ultrasonic detection part includes:

[0029] A first FPC layer is used for circuit connection, and a first electrode and a second electrode are led out from the same side of the first FPC layer to provide electrical energy.

[0030] A polymer material layer is arranged on one side of the first FPC layer and connected with the first electrode and the second electrode, for emitting or receiving ultrasonic waves to obtain the change of the blood vessel diameter;

[0031] A polymer material matching layer is arranged on the side of the polymer material layer away from the first FPC layer, for contacting the skin of the wrist;

[0032] The pulse wave detection part comprises:

[0033] A polymer substrate layer is arranged on the side of the first FPC layer away from the polymer material layer;

[0034] A piezoelectric ceramic is arranged on the side of the polymer substrate layer away from the first FPC layer, for detecting the pulse wave signal;

[0035] A second FPC layer is arranged on the side of the piezoelectric ceramic away from the polymer substrate layer, and a third electrode and a fourth electrode are led out on the side of the second FPC layer close to the piezoelectric ceramic, the third electrode and the fourth electrode are electrically connected with the piezoelectric ceramic, for providing electric energy for the piezoelectric ceramic.

[0036] Preferably, the ear PPG sensor comprises at least an optoelectronic PPG sensor;

[0037] The optoelectronic PPG sensor is used for emitting light of a specific wavelength to irradiate the ear tissue and receiving reflected light absorbed and reflected by the blood, converting the reflected light into a third electric signal, to obtain the ear PPG signal based on the third electric signal;

[0038] The ear ECG sensor detects the electrophysiological signal of the ear by contacting the skin of the ear, to obtain the ear ECG signal based on the electrophysiological signal of the ear.

[0039] Preferably, the LSTM model is constructed by the following steps:

[0040] A large amount of historical multi-point blood pressure data and physiological signal data of users are obtained to form a data set;

[0041] The data set is divided into a training set and a validation set;

[0042] Based on the training set, a neural network model is used for model training to output the mapping relationship / iteration logic / computational formula from the multi-point blood pressure data and the physiological signal data to the blood pressure prediction result, to form an initial prediction model;

[0043] Model verification is performed on the initial prediction model based on the verification set, so that the model prediction result reaches a preset accuracy by adjusting model parameters, thereby forming a final LSTM model.

[0044] Preferably, the smart watch and the two ear-hanging Bluetooth earphones communicate through Bluetooth mesh networking to realize real-time transmission and interaction of data.

[0045] The smart watch serves as a master node and is responsible for initiating and managing network connection, and the ear-hanging Bluetooth earphones serve as slave nodes and automatically join the network after detecting the network signal of the smart watch.

[0046] Preferably, the blood pressure feature is obtained through the following steps:

[0047] A digital filtering algorithm is used to remove noise and eliminate baseline drift for the wrist PPG signal, wrist ECG signal, ear PPG signal and ear ECG signal.

[0048] The pulse wave propagation time and pulse wave amplitude are extracted from the preprocessed wrist PPG signal and ear PPG signal, and the R wave peak value is extracted from the preprocessed wrist ECG signal and ear ECG signal.

[0049] The pulse wave propagation time, pulse wave amplitude and R wave peak value jointly constitute the blood pressure feature.

[0050] Preferably, the wrist PPG sensor includes a pressure film sensor and a PMUT sensor.

[0051] The pressure film sensor and the PMUT sensor are stacked on the watchband of the smart watch, or are arranged side by side on the watchband of the smart watch.

[0052] Preferably, the smart watch further comprises a storage module connected to the microprocessor, for storing historical blood pressure data for user viewing.

[0053] According to a second aspect of the embodiment of the present application, a multi-point non-invasive blood pressure monitoring method is provided, comprising:

[0054] Based on the smart wearable device worn on the specific tissue of the human body, a first detection signal is obtained through at least one first sensor.

[0055] Based on the ear-hanging Bluetooth earphone worn on the ears, a second detection signal is obtained through at least one second sensor.

[0056] transmitting the first detection signal and the second detection signal to a microprocessor in the intelligent wearable device through a communication module;

[0057] performing signal preprocessing and feature extraction on the first detection signal and the second detection signal through the microprocessor, forming a blood pressure feature after multi-site feature fusion;

[0058] taking the blood pressure feature as input, performing blood pressure calculation through a preset algorithm of the microprocessor, and outputting a blood pressure prediction result;

[0059] receiving and displaying the blood pressure prediction result in real time through a display screen on the intelligent wearable device.

[0060] Compared with the prior art, the application has the following technical effects:

[0061] (1) The embodiment of the application measures blood pressure by arranging sensors on the wrist and both ears at the same time, which provides multi-dimensional data compared with the traditional single-point measurement method. The signal at the wrist can reflect the condition of the wrist artery, while the signal at the ear can reflect the characteristics of the ear artery. By comprehensively analyzing these multi-site data, the blood pressure condition of the human body can be more comprehensively understood. Moreover, the blood pressure of each part is related to each other, thereby establishing an overall blood pressure measurement model, which can effectively improve the accuracy and reliability of blood pressure measurement.

[0062] (2) In the embodiment of the application, piezoelectric, pressure and photoelectric sensors are used in combination. The PPG signals obtained through multiple sensors can represent the volume changes of blood vessels and the propagation of pulse waves from multiple dimensions. Moreover, the ECG sensor is also used to collect ECG signals, thereby directly reflecting the electrical physiological activity of the heart. Therefore, by simultaneously collecting and analyzing multiple signals, more information about the cardiovascular system can be obtained, and the information collected by the sensors can also be verified and supplemented, thereby improving the accuracy and stability of measurement.

[0063] (3) In the embodiment of the application, the LSTM model is used in combination with the Kalman filtering algorithm to calculate the blood pressure prediction result. The LSTM model can extract and learn the features of the original data of multiple measurement points, mine the potential patterns behind the data, and integrate multi-source information. On this basis, the Kalman filtering algorithm dynamically adjusts the result of the data fusion of each measurement point according to the state transition matrix and the observation noise covariance matrix of the system. This combination can significantly improve the reliability of blood pressure measurement and provide more accurate data support for medical diagnosis.

[0064] (4) The smartwatch communicates with the two ear-hook Bluetooth headsets using mesh networking technology. The smartwatch acts as the master node, responsible for initiating and managing network connections, while the Bluetooth headsets act as slave nodes, automatically joining the network after detecting the smartwatch's network signal. Thus, the multi-hop communication characteristics of the mesh network enable relay transmission of signals between different nodes, expanding the communication range and improving communication reliability. Even if the signals of some nodes are interfered with or blocked, data can still be successfully transmitted to the smartwatch through other nodes.

[0065] (5) This embodiment of the invention integrates blood pressure monitoring functionality into smartwatches and Bluetooth headsets, enriching the functions of smart wearable devices. With increasing health awareness, the demand for smart wearable devices with health monitoring functions is constantly growing. This invention can meet consumers' daily health monitoring needs and has significant market potential. Simultaneously, the system's data storage and analysis functions can also provide users with health advice and exercise guidance, enhancing the user experience.

[0066] (6) In the field of healthcare, for patients with cardiovascular diseases who require long-term blood pressure monitoring, this invention can achieve real-time and accurate blood pressure monitoring, providing reliable data support for doctors to adjust treatment plans. For the elderly and patients with chronic diseases, the system's convenience and accuracy help them manage their daily health and improve their quality of life. Attached Figure Description

[0067] Figure 1 This is a structural diagram of a multi-point non-invasive blood pressure monitoring system provided in the first embodiment of the present invention;

[0068] Figure 2 This is a diagram showing the internal structure of the dial in the first embodiment of the present invention;

[0069] Figure 3 This is a schematic diagram of the layout of the wrist PPG sensor in the first embodiment of the present invention;

[0070] Figure 4 This is a schematic diagram of another layout of the wrist PPG sensor in the first embodiment of the present invention;

[0071] Figure 5A This is a schematic diagram of the structure of another wrist PPG sensor in the first embodiment of the present invention;

[0072] Figure 5B This is a schematic diagram illustrating the specific structural composition of another wrist-mounted PPG sensor in the first embodiment of the present invention;

[0073] Figure 6 The flowchart illustrates a multi-point non-invasive blood pressure monitoring method provided in the second embodiment of the present invention. DETAILED DESCRIPTION

[0074] The technical content of the present application will be described in detail below in combination with the drawings and specific embodiments.

[0075] The embodiments of the present application aim to solve the problems of insufficient precision, limited application range, and poor convenience of use of existing non-invasive blood pressure monitoring devices. Through innovative multi-point measurement methods (three-point measurement method of wrist + both ears in this embodiment), sensor fusion technology, and optimized data processing algorithms, more accurate, convenient, and widely applicable blood pressure measurement is achieved, providing a device that can monitor blood pressure in real time and accurately for users, meeting the needs of medical care, daily health management, and other aspects, especially providing a reliable blood pressure monitoring solution for people with cardiovascular diseases, people who need long-term blood pressure monitoring, and ordinary users who have high requirements for health management.

[0076] First embodiment

[0077] As shown in Figure 1 , the first embodiment of the present application provides a multi-point non-invasive blood pressure monitoring system, mainly composed of a smart watch 1 and two Bluetooth earphones 2. Among them, the smart watch 1 is the master control device of the system, with powerful data processing capability and rich function modules, including signal acquisition, data storage, display, communication, and algorithm operation, etc. The smart watch 1 is equipped with a high-performance microprocessor inside, which can quickly process data from various sensors and run complex blood pressure calculation algorithms. At the same time, the smart watch 1 is equipped with a high-resolution display screen, which is used to intuitively show the blood pressure measurement results, heart rate, time, and other important information to the user. In addition, the smart watch 1 also integrates a large-capacity storage module, which can be used to store historical blood pressure data, making it convenient for users to view and analyze their own blood pressure trends at any time.

[0078] The two Bluetooth earphones 2 are auxiliary measurement devices, respectively worn on both ears. The earphone is integrated with a small, low-power sensor module inside, which is used to collect physiological signals of the ear. This design fully utilizes the characteristics of rich blood vessels in the ear and convenient signal acquisition, and the hanging ear wearing method ensures the stability and comfort of the earphone during use, and it is not easy to fall off even when the user is performing daily activities. The Bluetooth earphone 2 communicates with the smart watch 1 through mesh networking technology, realizing real-time transmission and interaction of data.

[0079] Next, the software and hardware structure and working principle of the blood pressure monitoring system will be described in detail:

[0080] (I) Smart watch

[0081] As shown in Figure 1As shown, in this embodiment, the smart watch 1 is composed of a watch dial 11 and a watch band 12, and the watch dial 11 is worn on the wrist through the watch band 12. As shown in Figure 2 As shown, the watch dial 11 is internally provided with a microprocessor 111 and a communication module 112, wherein the microprocessor 111 is internally provided with a preset algorithm (to be described in detail below) for calculating a blood pressure prediction result according to blood pressure characteristics. The communication module 112 is connected with the microprocessor 111 for data transmission. In addition, the surface of the watch dial 11 is also provided with a display screen 113 connected with the microprocessor 111 for real-time display of the blood pressure prediction result. Furthermore, the internal part of the watch dial 11 is also provided with a storage module 114 connected with the microprocessor 111 for storing historical blood pressure data for the user to view.

[0082] As shown in Figure 1 As shown, in this embodiment, the watch band 12 can be any one of a woven band, a leather band or a chain band. The watch band 12 is provided with a wrist PPG sensor 121 and a wrist ECG sensor 122. When the watch dial 11 is worn on the wrist through the watch band 12, the wrist PPG sensor 121 and the wrist ECG sensor 122 will be correspondingly attached to the wrist skin, so as to acquire a wrist PPG (Photoplethysmography) signal by the wrist PPG sensor 121 and acquire a wrist ECG (Electrocardiography) signal by the wrist ECG sensor 122. In addition, the wrist PPG sensor 121 and the wrist ECG sensor 122 are both in communication connection with the communication module 112, so as to be able to transmit the wrist PPG signal and the wrist ECG signal to the microprocessor 111 through the communication module 112.

[0083] In this embodiment, preferably, the wrist PPG sensor 121 is a pressure film sensor and / or a PMUT sensor. It can be understood that when the heart beats, the blood vessels will produce periodic expansion and contraction, and the pressure film sensor converts the pressure change into a PPG signal by sensing the pressure change; the PMUT sensor emits and receives ultrasonic waves by using the piezoelectric effect, and detects the dynamic change of the blood vessels according to the propagation characteristics of the ultrasonic waves in the blood vessels to acquire a pulse wave.

[0084] In addition, the heart will produce weak electrical signals during the beating process, and these electrical signals will be conducted to the skin surface through the human body tissue, and the wrist ECG sensor 122 detects the electro-physiological signal of the wrist by contacting the wrist skin, so as to acquire the wrist ECG signal based on the electro-physiological signal of the wrist.

[0085] Specifically, the pressure film sensor is selected as a high-precision sensor based on the piezoelectric effect, such as a PVDF-TrFE pressure sensor. Such a sensor has high sensitivity and stability, and can accurately detect the slight pressure changes on the skin surface caused by pulse beats, convert them into electrical signals, and thus obtain high-quality PPG signals. The PMUT sensor is selected as LIS2DU12 of STMicroelectronics, which has the advantages of small size, low power consumption, and high sensitivity, and can accurately detect the dynamic changes of blood vessels by using the piezoelectric effect, and thus obtain accurate PPG signals. The wrist ECG sensor is selected as ADS1299 of Texas Instruments. This sensor is designed for bioelectric signal acquisition and has the characteristics of high input impedance, low noise, and high precision, and can accurately monitor the electrical physiological activity of the heart and obtain clear ECG signals.

[0086] In a preferred embodiment, the wrist PPG sensor 121 is composed of a pressure film sensor and a PMUT sensor. As shown in Figure 3 , the pressure film sensor and the PMUT sensor can be arranged in a stacked layout on the watchband 12 of the smart watch. As shown in Figure 4 , the pressure film sensor and the PMUT sensor can be arranged in a side-by-side layout on the watchband 12 of the smart watch. In addition, marks can be pre-set on the watchband 12 at positions corresponding to the sensors, so as to facilitate the user to align the sensors with the blood vessels and improve the detection effect.

[0087] In addition, as shown in Figure 5A , in another preferred embodiment, the wrist PPG sensor 121 includes an ultrasonic detection part 121a and a pulse wave detection part 121b. The ultrasonic detection part 121a is used to contact the human skin of a specific tissue, to emit or receive ultrasonic waves, and thus to obtain the diameter changes of the blood vessels. The pulse wave detection part 121b is arranged on the side of the ultrasonic detection part 121a away from the human skin, and is used to detect the pulse wave signals.

[0088] Specifically, referring to Figure 5BAs shown, the ultrasonic detection part 121a includes, from bottom to top, a polymer material matching layer 1211, a polymer material layer 1212, and a first FPC layer 1213. The polymer material matching layer 1211 is close to the skin and is used to contact the wrist skin. The polymer material layer 1212 is used to emit or receive ultrasonic waves to obtain the change of the blood vessel diameter. The first FPC layer 1213 is used to realize the overall circuit connection, and the same side (in this embodiment, the lower side) of the first FPC layer 1213 leads out a first electrode 1213a and a second electrode 1213b. It can be understood that the first electrode 1213a and the second electrode 1213b are positive and negative respectively, and can be interchanged, thereby jointly forming an electrode layer and being electrically connected with the polymer material layer 1212 to form a closed loop and provide electrical energy for the polymer material layer 1212.

[0089] Similarly, the pulse wave detection part 121b includes, from bottom to top, a polymer substrate layer 1214, a piezoelectric ceramic 1215, and a second FPC layer 1216. The polymer substrate layer 1214 is arranged on the side of the first FPC layer 1213 away from the polymer material layer 1212, so that the pulse wave detection part 121b is stacked on the ultrasonic detection part 121a to be away from the skin. The piezoelectric ceramic 1215 detects the pulse wave signal based on the piezoelectric effect. The second FPC layer 1216 is used to realize the overall circuit connection of the pulse wave detection part 121b, and the same side (in this embodiment, the lower side) of the second FPC layer 1216 leads out a third electrode 1216a and a fourth electrode 1216b. It can be understood that the third electrode 1216a and the fourth electrode 1216b are positive and negative respectively, and can be interchanged, thereby jointly forming an electrode layer and being electrically connected with the piezoelectric ceramic 1215 to form a closed loop and provide electrical energy for the piezoelectric ceramic 1215.

[0090] In this embodiment, the wrist PPG sensor 121 preferably further includes a reinforcing layer 1217. The reinforcing layer 1217 is arranged on the side of the second FPC layer 1216 away from the piezoelectric ceramic 1215 to provide support for the connection of the sensor.

[0091] It can be understood that the sensor structure in this embodiment continuously collects the pulse wave signal of the user by using the piezoelectric ceramic 1216, and continuously emits or collects ultrasonic wave signals by using the polymer material layer 1212 to obtain the change of the blood vessel diameter. It can be understood that the regularity of the pulse beat is basically consistent with the regularity of the change of the blood vessel diameter, and therefore the pulse wave signal collected by the piezoelectric ceramic 1216 can be corrected by the change data of the blood vessel diameter obtained by the polymer material layer 1212 to improve the accuracy of the PPG signal.

[0092] In addition, based on the special structure of the wrist PPG sensor 121, the manufacturing process is also different from the ordinary process flow, the specific differences are as follows:

[0093] (1) In this embodiment, the internal connection mode of the ultrasonic detection part 121a and the pulse wave detection part 121b is different.

[0094] Specifically, in the ultrasonic detection part 121a, the high polymer material layer 1212 and the first FPC layer 1213 are connected in the form of welding, and can also be connected in the form of wire bonding, but the bottom of the high polymer material layer 1212 needs to be filled with glue. In the pulse wave detection part 121b, the piezoelectric ceramic 1215 and the second FPC layer 1216 are bonded with non-conductive glue. And both steps need to be done in a vacuum environment to remove bubbles.

[0095] It should be noted that bubbles have a very large interference to ultrasonic signals, and removing bubbles can avoid signal interference. In this embodiment, on the one hand, by filling glue at the bottom of the high polymer material layer 1212, the high polymer material layer 1212 (for example: micro-electromechanical material) can be better connected with the first FPC layer 1213, so that the high polymer material matching layer 1211 as a whole acts as a backing to prevent ultrasonic energy from being transmitted to the back, so that the ultrasonic signal is concentrated to the front (i.e. the direction towards the pulse wave detection part 121b) to obtain better signal quality. On the other hand, based on experimental data, it is found that using non-conductive glue can make the ceramic have greater elastic deformation and thus better effect.

[0096] (2) The piezoelectric ceramic 1215 in this embodiment needs to be precisely ground to ensure that the flatness of the two surfaces meets the preset requirements, so as to apply pressure to the two layers of materials to combine them into a whole, and avoid the wafer material being crushed.

[0097] (II) Bluetooth earphone with ear hook

[0098] For example, Figure 1As shown, in the embodiment, two Bluetooth earphones 2 are respectively worn on the ears, and each Bluetooth earphone is internally provided with an ear PPG sensor 21, an ear ECG sensor 22, a Bluetooth module 23 and a master control module 24. Among them, when the Bluetooth earphone 2 is worn on the ear, the ear PPG sensor 21 will adhere to the ear skin, so as to obtain the ear PPG signal; and the ear ECG sensor 22 will adhere to the ear skin, so as to obtain the ear ECG signal. In addition, the ear PPG sensor 21 and the ear ECG sensor 22 are connected with the communication module 112 through the Bluetooth module 23 to realize mesh Bluetooth connection, so as to transmit the ear PPG signal and the ear ECG signal obtained from the two ears to the microprocessor 111 through the Bluetooth module 23 and the communication module 112. The master control module 24 is connected with the ear PPG sensor 21, the ear ECG sensor 22 and the Bluetooth module 23 respectively, and is used for controlling the operation of each component.

[0099] In the embodiment, the ear PPG sensor 21 is an optical PPG sensor. The optical PPG sensor can emit light of a specific wavelength to irradiate the ear tissue. Since the absorption and reflection characteristics of blood to light will change with the pulse, the optical PPG sensor receives the reflected light after being absorbed and reflected by the blood and converts it into a third electric signal, so as to obtain the ear PPG signal based on the third electric signal. The ear ECG sensor 22 detects the electrophysiological signal of the ear by contacting the ear skin, so as to obtain the ear ECG signal based on the electrophysiological signal of the ear.

[0100] In addition, in terms of communication, preferably, the Bluetooth module 23 and the communication module 112 both adopt Bluetooth chips supporting Bluetooth 5.0 and above standards. Bluetooth 5.0 has higher data transmission rate, longer transmission distance and lower power consumption, and can meet the real-time transmission requirements of a large amount of physiological signal data in the system. At the same time, it supports mesh networking technology, which can realize multi-hop communication between the smart watch and the Bluetooth earphone, and ensure the stability and reliability of data transmission.

[0101] In this embodiment, the smart watch 1 communicates with the two Bluetooth earphones 2 through Bluetooth mesh networking to realize real-time transmission and interaction of data. Mesh networking is a multi-node network topology based on Bluetooth Low Energy (BLE) technology, with characteristics such as self-organizing, self-repairing, and multi-hop communication. In this network, each device can act as a node, not only able to send and receive data, but also to forward data of other nodes, thereby realizing interconnection and intercommunication between devices. Specifically, the smart watch 1 acts as the master node, responsible for initiating and managing network connections; the Bluetooth earphones 2 act as slave nodes, which automatically join the network after detecting the network signal of the smart watch. When the Bluetooth earphones 2 collect physiological signals, they send the data to the smart watch 1 through the mesh network. Moreover, encryption and verification techniques can be used during data transmission to ensure data security and integrity. At the same time, the multi-hop communication feature of the mesh network enables signal relay transmission between different nodes, expanding the communication range and improving the reliability of communication. Even if some nodes are interfered or blocked, data can still be transmitted smoothly to the smart watch through other nodes. In addition, a smartphone can also be used as the master node.

[0102] (Three) Blood pressure monitoring principle

[0103] When the microprocessor 111 of the smart watch 1 receives PPG and ECG signals from the wrist and both ears, it needs to process the data through the preset algorithm built-in the microprocessor 111, so as to output the final blood pressure prediction result. In this embodiment, the preset algorithm calculates the blood pressure prediction result by combining a preset model with a Kalman weighting algorithm.

[0104] Specifically, the following steps are included:

[0105] (1) Data preprocessing

[0106] First, the PPG and ECG signals of the wrist and both ears are preprocessed, including noise removal, filtering, feature extraction, etc. By using advanced digital filtering algorithms such as Butterworth filter, Kalman filter, etc., high-frequency noise and baseline drift in the signal are removed, improving the quality of the signal.

[0107] Moreover, in this embodiment, the correspondence of multiple signals needs to be ensured during the preprocessing of the PPG and ECG signals of the wrist and both ears. That is, the signal data of the wrist and both ears are time-synchronized at the same time.

[0108] This is achieved by the following steps:

[0109] (1) Time synchronization technology

[0110] Hardware Synchronization: High-precision clock synchronization modules (such as GPS synchronized clocks or IEEE 1588 protocol devices) are used to ensure that the time references of the binaural and wrist signal acquisition devices are consistent, with an error controlled within microseconds.

[0111] Trigger Synchronization: External trigger signals (such as synchronization pulses) are used to simultaneously start data acquisition of each device, avoiding time drift. In this embodiment, multiple points can measure ECG signals, and each point uses the R wave of the ECG signal as a synchronization signal.

[0112] Software Marking: Time stamps (such as NTP protocol or custom time markers) are embedded in the data stream, which are later aligned by time stamps.

[0113] (2) Signal Alignment and Association

[0114] Feature Point Matching: Feature points in the signal (such as the starting point and peak value of the pulse wave) are used as anchor points, and algorithms (such as dynamic time warping, DTW) are used to align signals from different parts.

[0115] Cross-Validation: The accuracy of alignment is verified by the similarity of multi-site signals (such as pulse wave period and shape), and abnormal signals are removed.

[0116] Multi-modal Fusion Algorithm: Deep learning models (such as LSTM or Transformer) are used to process multi-site signals, automatically learning the spatio-temporal relationship between signals.

[0117] (3) Data Preprocessing and Correction

[0118] Filtering and Noise Reduction: Binaural and wrist signals are filtered (such as band-pass filter) respectively to eliminate environmental noise and device noise.

[0119] Baseline Correction: Remove the direct current offset in the signal to ensure consistent baseline.

[0120] Delay Compensation: Transmission delays of signals from different parts are measured through experiments, and compensation is made during data processing stage.

[0121] (2) Feature Extraction

[0122] Feature extraction algorithms are used to extract blood pressure-related feature parameters from PPG and ECG signals, such as pulse wave transmission time (PTT), pulse wave amplitude, and R wave peak value of ECG signal. Bluetooth earphones can also process ear PPG and ear ECG signals locally, and only send blood pressure-related features to microprocessor 111.

[0123] In addition, in this embodiment, after the extraction of blood pressure-related features, feature fusion is also needed as the input of the preset model

[0124] (3) outputting the blood pressure prediction result

[0125] In this embodiment, a large amount of historical blood pressure data and corresponding physiological signal data are trained by a machine learning algorithm such as a support vector machine (SVM) or a neural network to construct an LSTM model (i.e., a preset model) for predicting blood pressure. Thus, after feature fusion is completed, the fused features of multiple parts are input into the pre-trained LSTM model, and a blood pressure prediction result is output based on the LSTM model.

[0126] Specifically, in this embodiment, the ECG provides a heartbeat synchronization reference for calibrating the time points of other signals, the vascular diameter change rate reflects the vascular elasticity, and the multi-site signals (wrist + ear) can improve the signal reliability. First, each data is aligned to the R wave of the ECG. Then, a band-pass filter is used to remove baseline drift and high-frequency noise. The features of each group of signals are arranged in time synchronization to form a multi-dimensional vector. An attention layer is then introduced to dynamically allocate weights to each group of signals.

[0127] The LSTM model is constructed by the following steps:

[0128] ① A large amount of multi-point blood pressure data and physiological signal data of historical users are obtained to form a data set;

[0129] ② The data set is divided into a training set and a validation set;

[0130] ③ Based on the training set, a neural network model is used for model training, a sliding window is used to generate time series samples, each batch contains multiple time series, and a mapping relationship / iteration logic / computational formula from multi-point blood pressure data and physiological signal data to blood pressure prediction result is output, thereby forming an initial prediction model;

[0131] ④ The initial prediction model is verified based on the validation set, the loss function of the validation set is monitored, and if there is no improvement for 15 consecutive rounds, the training is stopped, the model parameters are adjusted to reduce the output error, the model prediction result reaches the preset accuracy, and the final LSTM model is formed.

[0132] The loss function of the LSTM model uses mean square error (MSE). The training is completed under the following conditions:

[0133] 1. The loss function converges and is stable;

[0134] 2. When the loss or prediction accuracy does not improve (even decreases) in a plurality of iterations (such as 15 iterations), the training is terminated in advance;

[0135] 3. The prediction error reaches the clinically acceptable range;

[0136] 4. Set the maximum number of iterations (such as 200 rounds) as the model training termination condition.

[0137] In addition, preferably, during the training of the LSTM model, parameters such as gender, age, and body fat rate can be added as auxiliary inputs to improve the generalization of the model.

[0138] In addition, in the present embodiment, the Kalman weighting algorithm is embedded in the preset model, which is used to assign different weights to the wrist PPG signal and the ear PPG signal in real time according to the preset logic, so as to correct the blood pressure prediction result in real time. Specifically, in the algorithm, different weights are assigned to the data of each part according to factors such as the accuracy, stability and individual difference of different part sensors. For example, for the data at the wrist, a relatively low weight is assigned because it is relatively convenient to measure but is greatly disturbed by motion; while for the data at the ear, a relatively high weight is assigned because it is less disturbed by the outside world and the signal is relatively stable.

[0139] Specifically, the following logic is used to assign weights in the present embodiment:

[0140] (1) According to "part" division: limbs > ear, right side > left side (default)

[0141] The core logic is that there is a "decay gradient" from the heart to the peripheral blood vessels, and the limbs (such as the arms) close to the heart can better reflect the central blood pressure (high consistency with standard blood pressure); the ear blood vessels are thin and belong to peripheral circulation, which is greatly affected by peripheral resistance, and the signal stability is poor.

[0142] The specific part weight reference (total 100%) is as follows:

[0143] ① Right hand (arm / wrist): 30-40%. The right upper arm is traditionally considered the "gold standard" for blood pressure measurement,

[0144] which is close to the heart level and has the highest correlation with SBP / DBP (R 2 ≈0.7-0.8).

[0145] ② Left hand (arm / wrist): 25-30%. The difference with the right hand is small (usually <5 mmHg), but the right hand is preferred in clinical practice, so the weight is slightly lower.

[0146] ③ Right ear: 15-20%. Peripheral vascular signal, reflecting peripheral resistance, auxiliary supplementary information, but greatly affected by emotions and temperature (low SNR).

[0147] ④ Left ear: 10-15%. Symmetrical with the right ear, the signal stability is slightly lower than the right ear (the blood flow of the left ear of some people is weaker), and the weight is slightly lower.

[0148] (2) According to "signal type": PPG / pulse wave signal > ECG signal

[0149] Core logic: Blood pressure is essentially the pressure of blood on the vessel wall, and PPG and pulse wave directly reflect the changes in hemodynamics (such as pulse wave conduction velocity PWV, amplitude changes), and are more directly related to blood pressure; ECG mainly provides the "time reference" of cardiac electrical activity (such as R wave for calculating PWV), and indirectly assists in blood pressure evaluation.

[0150] Specific signal type weight reference (total 100%) as follows:

[0151] ① PPG / pulse wave: 60-70%. Directly reflects vascular elasticity, blood flow velocity, and has strong correlation with SBP (systolic blood pressure) (PWV↑→SBP↑).

[0152] ② ECG: 30-40%. Provides heart rate, R wave time point (used for calculating PWV = distance / time difference), and assists in calibration, but has weak ability to predict blood pressure alone.

[0153] (3) Comprehensive weight: cross distribution of parts x signal type

[0154] In this embodiment, the weights of "parts" and "signal types" need to be combined (product method can be used), for example:

[0155]

[0156] On this basis, by reasonably adjusting the weights, the signal data of the three parts can be integrated to help the LSTM model more accurately predict the current blood pressure of the human body.

[0157] It is worth noting that blood pressure data exhibits dynamic characteristics within a day, and the LSTM network can capture long-term dependencies in the data. Its special gating mechanism (input gate, forget gate, and output gate) allows the model to selectively remember important information from past time steps and forget irrelevant information. In multi-point blood pressure monitoring, there is an inherent relationship between blood pressure data at different times and different measurement points. The LSTM model can learn these rules well, thus accurately predicting and analyzing blood pressure trends. For example, for the slow changes in blood pressure during sleep at night, the LSTM model can accurately predict the blood pressure value at the next time point by learning the data at multiple previous time points. Using the system's state equation and observation equation, combined with the statistical characteristics of noise, the predicted value is optimally estimated to remove noise interference in the data, making the blood pressure data smoother and more accurate, effectively reflecting the true dynamic changes in blood pressure. Multi-point blood pressure measurement aims to obtain more comprehensive and accurate blood pressure information through measurements at multiple sites. However, due to individual differences (such as different skin colors and ages leading to different tissue characteristics) and measurement environment factors, there may be noise and errors in the data from each measurement point. LSTM can extract features and learn from the original data from multiple measurement points, uncovering the underlying patterns in the data and preliminarily integrating multi-source information.

[0158] The Kalman filter algorithm dynamically adjusts the results of data fusion from each measurement point based on the system's state transition matrix and observation noise covariance matrix. It can continuously update the estimate of the true blood pressure value based on the latest measurement value and the previous estimate, effectively reducing the impact of measurement noise and improving the accuracy and stability of blood pressure measurement. Compared to models that use a single measurement point or only one algorithm, this combined approach can significantly improve the reliability of blood pressure measurement and provide more accurate data support for medical diagnosis.

[0159] Different individuals have different physiological characteristics of blood pressure, such as different blood pressure fluctuation patterns between hypertensive patients and normal people, and different blood pressure change patterns between the elderly and young people. The LSTM model has strong non-linear modeling capabilities and can be trained on a large amount of multi-point blood pressure data from different individuals to learn their unique blood pressure change patterns. For each individual's multi-point blood pressure data, LSTM can build a personalized model to accurately capture their blood pressure characteristics. The Kalman filter algorithm can dynamically adjust the model parameters based on the individual's real-time blood pressure measurement data, making the model better adapt to individual changes. For example, when an individual's physical state changes (such as exercise or emotional fluctuations) causing abnormal changes in blood pressure, the Kalman filter can adjust the estimate in a timely manner, allowing the model to always accurately track the individual's blood pressure, thus providing more accurate blood pressure monitoring and analysis results for different individuals.

[0160] (Four) Software development and implementation of the monitoring system

[0161] In this embodiment, the development of system software adopts a layered architecture design, mainly including the driver layer, middleware layer and application layer.

[0162] The driver layer is responsible for interacting with hardware devices, implementing control and data collection of sensors, communication modules and other hardware. For different hardware devices, develop corresponding drivers. For example, write data collection drivers for pressure film sensors, PMUT sensors, ECG sensors, etc. to realize the initialization, configuration and data reading operations of these sensors. At the same time, develop Bluetooth communication drivers to realize communication control with Bluetooth modules, including functions such as Bluetooth device search, connection, data sending and receiving, etc.

[0163] The middleware layer mainly realizes signal processing, data management and communication protocol functions. In terms of signal processing, digital filtering algorithms are used to preprocess the collected PPG and ECG signals. For example, use FIR low-pass filter to remove high-frequency noise in the signal, and smooth the signal to improve the quality of the signal. Feature extraction algorithms are used to extract blood pressure-related feature parameters from the preprocessed signals, such as pulse wave propagation time (PTT), pulse wave amplitude, R-wave peak value of ECG signal, etc.

[0164] In terms of data management, data storage and management functions are realized. The collected physiological signal data and calculated blood pressure data are stored in the local storage of the smart watch, and data synchronization to the cloud server is supported, so that users can view and analyze through mobile phone APP or web page. In terms of communication protocol, a communication protocol based on Bluetooth mesh is implemented to ensure safe, reliable and efficient data transmission between the smart watch and the Bluetooth earphone. The transmitted data is encrypted to prevent data theft or tampering; data verification mechanism is adopted to ensure data integrity.

[0165] The application layer mainly realizes user interface and blood pressure calculation functions. In terms of user interface, a simple and intuitive user interface is developed to facilitate user operation and viewing of measurement results. The user interface displays real-time blood pressure, heart rate, pulse wave and other information, and provides historical data query, health advice and other functions. In terms of blood pressure calculation, weighted average algorithm is implemented to calculate the preliminary blood pressure value according to the data collected by different part sensors, combined with the reliability and stability of each part data, and assign appropriate weights to them. Through weighted average, the preliminary blood pressure value is calculated. Then, combined with machine learning algorithms such as neural networks, a large amount of blood pressure data and corresponding physiological signal data are trained to establish a personalized blood pressure prediction model, further optimize the blood pressure calculation result, and improve the accuracy of blood pressure measurement.

[0166] Second embodiment

[0167] As Figure 6As shown, on the basis of the above multi-point non-invasive blood pressure monitoring system, the second embodiment of the present application also provides a multi-point non-invasive blood pressure monitoring method, which specifically comprises the following steps:

[0168] S10: Obtain multi-point signal data.

[0169] Specifically, based on the smart watch 1 worn on the wrist, the wrist PPG signal is obtained through the wrist PPG sensor 121, and the wrist ECG signal is obtained through the wrist ECG sensor 122. Correspondingly, based on the ear-hanging Bluetooth earphone 2 worn on both ears, the ear PPG signal is obtained through the ear PPG sensor 21, and the ear ECG signal is obtained through the ear ECG sensor 22. Wherein, the signal acquisition method of each sensor can refer to the description in the above blood pressure monitoring system, which will not be repeated here.

[0170] And when the multi-point signal data is obtained through each sensor, the multi-point signal data needs to be transmitted to the microprocessor. Specifically, based on mesh networking, the communication module 112 sends a signal acquisition demand, and the wrist PPG sensor 121 and the wrist ECG sensor 122 respectively transmit the wrist PPG signal and the wrist ECG signal to the microprocessor of the smart watch. And the Bluetooth module 23 of the two ear-hanging Bluetooth earphones 2 automatically joins the network after detecting the network signal of the smart watch, so as to transmit the ear PPG signal and the ear ECG signal to the microprocessor of the smart watch through the communication module.

[0171] S20: Data processing.

[0172] Specifically, when the microprocessor obtains the multi-point signal data, the multi-point signal data needs to be preprocessed and feature extracted, so as to form the blood pressure feature after multi-site feature fusion. Wherein, the data preprocessing and feature extraction process can refer to the description in the above blood pressure monitoring system, which will not be repeated here.

[0173] S30: Output blood pressure prediction result.

[0174] Specifically, this step is divided into two parts, one part is to output the blood pressure prediction result according to the blood pressure feature after multi-site feature fusion by using the pre-constructed LSTM model; the other part is to give different weights to the wrist PPG signal and the two ear PPG signals by Kalman weighting algorithm before the LSTM model outputs the blood pressure prediction result, so as to real-time correct the blood pressure prediction result, and finally output the corrected true blood pressure value.

[0175] S40: Real-time receive and display blood pressure prediction result through display screen.

[0176] In summary, the multi-point non-invasive blood pressure monitoring system and method provided by the embodiments of the present application has the following beneficial effects:

[0177] (1) The embodiments of the present application measure blood pressure by arranging sensors on the wrist and both ears at the same time, which provides multi-dimensional data compared with the traditional single-point measurement method. The signal at the wrist can reflect the condition of the wrist artery, and the signal at the ear can reflect the characteristics of the ear artery. By comprehensively analyzing the data of these multiple parts, the blood pressure condition of the human body can be more comprehensively understood, and the blood pressure of each part is related to each other, thereby establishing a whole blood pressure measurement model, which can effectively improve the accuracy and reliability of blood pressure measurement.

[0178] (2) In the embodiments of the present application, piezoelectric, pressure and photoelectric sensors are used in combination, and the PPG signals obtained by multiple sensors can represent the volume changes of blood vessels and the propagation of pulse waves from multiple dimensions. Moreover, the ECG signal is collected in cooperation with the ECG sensor, thereby directly reflecting the electrical physiological activity of the heart. Therefore, by simultaneously collecting and analyzing multiple signals, more information about the cardiovascular system can be obtained, and the information collected by the sensors can be verified and supplemented with each other, thereby improving the accuracy and stability of measurement.

[0179] (3) In the embodiments of the present application, the LSTM model is used in combination with the Kalman filtering algorithm to calculate the blood pressure prediction result, the LSTM model can extract and learn the features of the original data of multiple measurement points, mine the potential patterns behind the data, and integrate multi-source information. On this basis, the Kalman filtering algorithm dynamically adjusts the result of the data fusion of each measurement point according to the state transition matrix and the observation noise covariance matrix of the system. This combination can significantly improve the reliability of blood pressure measurement and provide more accurate data support for medical diagnosis.

[0180] (4) The mesh networking technology is used for communication between the smart watch and the two Bluetooth earphones, the smart watch is used as the master node to initiate and manage network connection, and the Bluetooth earphones are used as the slave nodes to automatically join the network after detecting the network signal of the smart watch. Therefore, based on the multi-hop communication characteristics of the mesh network, the signal can be relayed between different nodes, which expands the communication range and improves the reliability of communication. Even in the case that part of the nodes are interfered or blocked, the data can still be transmitted to the smart watch through other nodes.

[0181] (5) The embodiment of the present application integrates blood pressure monitoring function into smart watches and Bluetooth earphones, enriching the functions of smart wearable devices. With the improvement of people's health awareness, the demand for smart wearable devices with health monitoring function is increasing. The product of the present application can meet the needs of consumers for daily health monitoring, and has great market potential. At the same time, the data storage and analysis function of the system can also provide health advice and exercise guidance for users, improving user experience.

[0182] (6) In the field of medical care, for patients with cardiovascular diseases who need long-term monitoring of blood pressure, the present application can realize real-time and accurate blood pressure monitoring, providing reliable data support for doctors to adjust treatment plans. For the elderly and chronic disease patients, the convenience and accuracy of the system help them to manage their daily health and improve their quality of life.

[0183] It should be noted that the above multiple embodiments are only illustrative. The technical solutions of each embodiment can be combined, and all are within the protection scope of the present application.

[0184] It should be noted that the terms "thickness", "depth", "upper", "lower", "horizontal" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0185] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0186] The multi-point non-invasive blood pressure monitoring system and method provided by the present application are described in detail above. Any obvious modification made by a person skilled in the art without departing from the essential content of the present application will constitute an infringement of the patent right of the present application, and the person skilled in the art will bear the corresponding legal responsibility.

Claims

1. A multi-point non-invasive blood pressure monitoring system, characterized by, The system comprises a smart wearable device and a Bluetooth earphone, wherein the smart wearable device is communicatively connected with the Bluetooth earphone. The smart wearable device is worn on a specific tissue of a human body and adheres to the skin of the human body, and at least one first sensor is arranged on the smart wearable device to acquire a first detection signal of a region where the specific tissue is located. The Bluetooth earphone is worn on both ears, and at least one second sensor is arranged on the Bluetooth earphone to acquire a second detection signal of the ears. The smart wearable device further comprises a microprocessor and a communication module, and the microprocessor is communicatively connected with the first sensor and the second sensor through the communication module to receive the first detection signal and the second detection signal, perform multi-site feature fusion, form a blood pressure feature, and calculate a blood pressure prediction result based on a preset algorithm according to the blood pressure feature.

2. The blood pressure monitoring system of claim 1, wherein: the smart wearable device is a smart watch worn on a wrist, and the at least one first sensor comprises a wrist PPG sensor and a wrist ECG sensor; the smart watch further comprises a display screen connected with the microprocessor to display the blood pressure prediction result; the wrist PPG sensor adheres to the skin of the wrist to acquire a wrist PPG signal; the wrist ECG sensor adheres to the skin of the wrist to acquire a wrist ECG signal; and the wrist PPG sensor and the wrist ECG sensor are both communicatively connected with the communication module; the at least one second sensor comprises an ear PPG sensor and an ear ECG sensor, the ear PPG sensor adheres to the skin of the ear to acquire an ear PPG signal, and the ear ECG sensor adheres to the skin of the ear to acquire an ear ECG signal; and the ear PPG sensor and the ear ECG sensor are both communicatively connected with the communication module; the microprocessor receives the wrist PPG signal, the wrist ECG signal, the ear PPG signal, and the ear ECG signal through the communication module, performs multi-site feature fusion after signal preprocessing and feature extraction, forms a blood pressure feature, and calculates a blood pressure prediction result by using a preset model and a Kalman weighting algorithm as input, and displays the blood pressure prediction result in real time through the display screen; the preset model adopts a neural network, a large amount of historical blood pressure data and corresponding physiological signal data are used for model training to construct an LSTM model for predicting blood pressure; and the LSTM model is used to output a blood pressure prediction result according to the wrist PPG signal, the ear PPG signal, the wrist ECG signal, and the ear ECG signal; the Kalman weighting algorithm is embedded in the preset model, and is used to real-time assign different weights to the wrist PPG signal and the ear PPG signal according to a preset logic, so as to real-time correct the blood pressure prediction result. 3.The blood pressure monitoring system of claim 2, wherein: the wrist PPG sensor comprises at least a pressure film sensor and / or a PMUT sensor; the pressure film sensor is configured to acquire a slight pressure change of the wrist skin surface caused by pulse beat and convert the pressure change into a first electrical signal, so as to acquire a wrist PPG signal based on the first electrical signal; the PMUT sensor is configured to acquire an ultrasonic wave change of the wrist skin surface caused by pulse beat and convert the ultrasonic wave change into a second electrical signal, so as to acquire a wrist PPG signal based on the second electrical signal; the wrist ECG sensor is configured to detect an electro-physiological signal of the wrist by contacting the wrist skin, so as to acquire a wrist ECG signal based on the electro-physiological signal of the wrist.

4. The blood pressure monitoring system of claim 1, wherein the first sensor comprises: an ultrasonic detection portion configured to contact a human skin of a specific tissue, for emitting or receiving ultrasonic waves, so as to acquire a blood vessel diameter change; a pulse wave detection portion disposed on a side of the ultrasonic detection portion away from the human skin, for detecting a pulse wave signal; wherein, at the same time, the blood vessel diameter change matches the pulse wave signal, so as to correct the pulse wave signal based on the blood vessel diameter change. 5.The blood pressure monitoring system of claim 4, wherein: the ultrasonic detection portion comprises: a first FPC layer for circuit connection, and a first electrode and a second electrode are led out from a same side of the first FPC layer for providing electrical energy; a high polymer material layer disposed on a side of the first FPC layer and connected with the first electrode and the second electrode, for emitting or receiving ultrasonic waves, so as to acquire a blood vessel diameter change; a high polymer material matching layer disposed on a side of the high polymer material layer away from the first FPC layer, for contacting the wrist skin; the pulse wave detection portion comprises: a high polymer substrate layer disposed on a side of the first FPC layer away from the high polymer material layer; a piezoelectric ceramic disposed on a side of the high polymer substrate layer away from the first FPC layer, for detecting a pulse wave signal; a second FPC layer disposed on a side of the piezoelectric ceramic away from the high polymer substrate layer, and a third electrode and a fourth electrode are led out from a side of the second FPC layer close to the piezoelectric ceramic, the third electrode and the fourth electrode are electrically connected with the piezoelectric ceramic for providing electrical energy for the piezoelectric ceramic. 6.The blood pressure monitoring system of claim 2, wherein: the ear PPG sensor comprises at least a photoelectric PPG sensor; the photoelectric PPG sensor is configured to emit light of a specific wavelength to irradiate an ear tissue and receive reflected light absorbed and reflected by blood, and convert the reflected light into a third electrical signal, so as to acquire an ear PPG signal based on the third electrical signal; the ear ECG sensor is configured to detect an electro-physiological signal of the ear by contacting the ear skin, so as to acquire an ear ECG signal based on the electro-physiological signal of the ear.

7. The blood pressure monitoring system of claim 2, wherein the LSTM model is constructed by the following steps: Obtain a large amount of historical multi-point blood pressure data and physiological signal data of users to form a data set; Divide the data set into a training set and a validation set; Based on the training set, perform model training through a neural network model to output a mapping relationship / iterative logic / computational formula from multi-point blood pressure data and physiological signal data to blood pressure prediction results, thereby forming an initial prediction model; Based on the validation set, perform model verification on the initial prediction model to adjust model parameters so that the model prediction result reaches a preset accuracy, thereby forming a final LSTM model.

8. The blood pressure monitoring system of claim 2, wherein: the smart watch and the two over-ear Bluetooth earphones communicate through Bluetooth mesh networking to realize real-time transmission and interaction of data; the smart watch serves as a master node and is responsible for initiating and managing network connection; and the over-ear Bluetooth earphones serve as slave nodes and automatically join the network after detecting the network signal of the smart watch. The blood pressure features are obtained through the following steps: For the wrist PPG signal, wrist ECG signal, ear PPG signal, and ear ECG signal, a digital filtering algorithm is used to remove noise and eliminate baseline drift; 9. The blood pressure monitoring system of claim 2, wherein The pulse wave propagation time and pulse wave amplitude are extracted from the preprocessed wrist PPG signal and ear PPG signal; and the R-wave peak value is extracted from the preprocessed wrist ECG signal and ear ECG signal; The pulse wave propagation time, pulse wave amplitude, and R-wave peak value jointly constitute the blood pressure features.

10. The blood pressure monitoring system of claim 2, wherein: The wrist PPG sensor includes a pressure film sensor and a PMUT sensor; The pressure film sensor and the PMUT sensor are stacked on the watchband of the smart watch, or are arranged side by side on the watchband of the smart watch.

11. The blood pressure monitoring system of claim 2, wherein: The internal of the smart watch is further provided with a storage module connected with the microprocessor, for storing historical blood pressure data for user to view. The method includes the following steps: Based on the smart wearable device worn on the specific tissue of the human body, a first detection signal is obtained through at least one first sensor; 12. A method of multi-point non-invasive blood pressure monitoring, characterized by Based on the over-ear Bluetooth earphones worn on the ears, a second detection signal is obtained through at least one second sensor; The first detection signal and the second detection signal are transmitted to the microprocessor of the smart wearable device through a communication module; The first detection signal and the second detection signal are preprocessed and feature-extracted by the microprocessor, and then the blood pressure features are formed after multi-site feature fusion; The blood pressure features are taken as input, and the blood pressure is calculated through a preset algorithm of the microprocessor, thereby outputting the blood pressure prediction result; The blood pressure prediction result is received and displayed in real time through the display screen on the smart wearable device. ​ ​

Images