CN-122004925-A - Multi-arterial coupling wearable continuous blood pressure measuring device and method based on multi-modal sensing

Abstract

The invention discloses a multi-arterial coupling wearable continuous blood pressure measuring device and method based on multi-modal sensing, and relates to the technical field of medical monitoring. Existing wearable devices also have inherent limitations in hardware configuration, which are generally limited to acquiring peripheral signals at a single location, and such sensor arrangements are not capable of acquiring multi-modal physiological signals distributed at different anatomical locations necessary for constructing a high-fidelity hemodynamic model. A first object of the present invention is to provide a blood pressure measuring device with a specific sensor arrangement specifically designed for simultaneous acquisition of central arterial blood flow information and peripheral arterial pulse signals required for implementing the above method, thereby providing hardware support for high-precision personalized blood pressure estimation. The second objective is to provide a blood pressure measurement method, which is based on 0D and 1D hemodynamic models, can fuse central and external Zhou Duomo-state signals, and overcomes the defects of the existing algorithm.

Inventors

• DING XIAORONG
• LIN YUXUAN

Assignees

• 电子科技大学

Dates

Publication Date

20260512

Application Date

20260113

Claims (5)

1. 1. A multi-arterial coupling wearable continuous blood pressure measuring device based on multi-modal sensing comprises a central arterial signal acquisition link, a peripheral arterial signal acquisition link, a main controller and a processor module; The central artery signal acquisition link comprises a central artery signal acquisition module, a first signal conditioning module, a first analog-to-digital conversion module and a first data processing module; Collecting an original simulated central artery blood flow signal by a central artery signal collecting module, regulating the original simulated central artery blood flow signal by a first signal regulating module, converting the regulated simulated central artery blood flow signal into a digital central artery blood flow signal by a first analog-to-digital conversion module, and preprocessing the digital central artery blood flow signal by a first data processing module; the peripheral arterial signal acquisition link comprises a peripheral arterial signal acquisition module and a second data processing module; Adopting a peripheral arterial signal acquisition module to acquire an original peripheral arterial pulse signal, and adopting a second data processing module to process, package or convert the original peripheral arterial pulse signal; the main controller receives the total control signal from the processor module and sends control signals to the central artery signal acquisition module and the peripheral artery signal acquisition module to realize synchronous acquisition; The processor module sends a total control signal to the controller module to control the opening or closing of the measuring program, receives the central artery blood flow information and the peripheral artery pulse signals from the first data processing module and the second data processing module respectively, stores the data, inputs the stored data into an artery 0D-1D coupling model, and the model comprises a preprocessing unit and a blood pressure resolving unit, calculates blood pressure results and outputs the blood pressure results to the display unit.
2. 2. A method of using the multi-modal sensing based multi-arterial coupled wearable continuous blood pressure measurement device of claim 1, the method comprising the steps of: Step 1, signal acquisition and conditioning; According to a user control instruction, a measurement program is started, the central artery signal acquisition module and the peripheral artery signal acquisition module are controlled to synchronously acquire, and respectively output an original simulated central artery blood flow signal and an original simulated peripheral artery pulse signal, wherein the original simulated central artery blood flow signal and the original simulated peripheral artery pulse signal are collectively called as original analog signals, and meanwhile, a reference continuous blood pressure waveform acquisition device is controlled to be started to acquire a reference blood pressure waveform; Step 2, signal conversion and transmission; The conditioned analog signals are transmitted to an analog-to-digital conversion module, and the analog-to-digital conversion module converts the conditioned original analog signals into digital signals; the signal transmission module transmits the original converted digital signal to the processor module; Step 3, signal preprocessing; the pretreatment unit in the processor module receives and analyzes the transmitted digital signals, and the pretreatment unit carries out digital filtering and denoising treatment on the analyzed digital signals so as to obtain clear central artery blood flow information and reference peripheral artery pulse signals; step 4, establishing a coupling model in the processor module, wherein the coupling model is configured to characterize a blood flow path from a central artery to a peripheral artery, and the coupling model comprises: s1, central artery modeling part: The central artery modeling portion is a total parameter model configured to simulate the overall hemodynamic characteristics at the central artery and its common node, the central artery modeling portion is an equivalent electrical network model in which hemodynamic parameters are analogized to circuit parameters and blood pressure is analogized to voltage The blood flow rate is analogized to current The blood flow resistance of a blood vessel is analogized to an electrical resistance , Wherein In order to be a viscosity of the blood, And Respectively the first The length and radius of the individual vessel segments; Arterial compliance is analogized to capacitance Its value is based on the geometric parameters of the blood vessel 、 And vessel wall material characteristics, the vessel wall material characteristics are Young's modulus E and wall thickness h, and the blood inertia is analogized to inductance , Calculation of wherein Is the blood density; S2, a target blood pressure measuring section artery modeling part: The target blood pressure measurement section artery modeling part is a distribution parameter model, and is configured to simulate one-dimensional pulse wave propagation characteristics of blood in the target blood pressure measurement section artery, the target blood pressure measurement section artery modeling part is constructed based on a fluid dynamics physical law, the physical law is one or more groups of partial differential equations, and the partial differential equations comprise: Mass conservation equation: ; Momentum conservation equation: ; Wherein, the Is the cross-sectional area of the blood vessel, For the blood flow rate, In order to be a blood pressure level, For the location along the blood vessel, In order to be able to take time, As a function of the velocity of the blood flow, Is the blood density; Is a non-linear velocity profile coefficient, Is a coefficient related to friction; Step 5, calculating blood pressure; the blood pressure calculating unit in the processor module receives the central artery blood flow information and the reference peripheral artery pulse signals processed by the preprocessing unit in the step 3, and executes the following substeps by using the model defined in the step 4: Step 5.1, determining inlet boundary conditions: The blood pressure calculation unit processes the central artery blood flow information and applies the central artery modeling portion defined in step 4 to determine one or more sets of hemodynamic parameters that characterize the inlet status of the target artery; the determined hemodynamic parameters are used to establish inlet boundary conditions for the arterial modeling portion of the target blood pressure measurement segment defined in step 4; step 5.2, performing hemodynamic personalized modeling: The blood pressure calculation unit performs an optimization process using one or more data constraints, which may include the inlet boundary conditions determined in sub-step 1 and the reference peripheral signal obtained in step 3; the optimization process is to optimize a group of subject specific parameters in the arterial modeling part of the target blood pressure measuring section, wherein the specific parameters comprise Young modulus of the vascular wall Wall thickness of blood vessel Or coefficient of friction ; And 5.3, calculating a blood pressure waveform and a blood pressure value, wherein the blood pressure calculating unit determines the blood pressure value of the target artery by using the specific parameters obtained in the substep 5.2, reconstructing a continuous blood pressure waveform of the brachial artery in one or more cardiac cycles, and extracting the systolic pressure and the diastolic pressure in the continuous blood pressure waveform.
3. 3. A method of a multi-modal sensing based multi-arterial coupled wearable continuous blood pressure measurement device as claimed in claim 2, wherein the optimization procedure in step 5.2 comprises the specific steps of: Step 5.2.1, parameter initialization and iteration, namely initializing specific parameter population # , , ) And iterating the subject specific parameters using a genetic algorithm; In each iteration, using the parameters of the current iteration, solving the equation set of the coupling model through a numerical value solver to obtain an estimated peripheral pulse waveform; And 5.2.3, calculating and updating errors, namely calculating the characteristic difference between the estimated peripheral pulse waveform and the reference peripheral pulse signal to be used as an estimated error, and updating parameters according to the errors by a genetic algorithm until the errors are converged.
4. 4. A method of a multi-modal sensing based multi-arterial coupled wearable continuous blood pressure measurement device as claimed in claim 2, wherein the optimization procedure in step 5.2 comprises the specific steps of: S5.2.1. establishing an estimate generator G1: The G1 is a neural network configured to receive as inputs the central arterial blood flow information and the reference peripheral signal; the output of the G1 is an estimated target arterial hemodynamic waveform, which comprises an estimated brachial artery blood pressure waveform and an estimated brachial artery blood flow waveform; s5.2.2 building a physical simulation generator G2: The G2 is a physical information neural network configured to receive the estimated target arterial hemodynamic waveform of the G1 output as its inlet boundary condition, the output of the G2 including blood pressure and blood flow waveforms of intermediate nodes on the 1D model path; s5.2.3 two discriminators are set up, D1 and D2: A first discriminator D1 is configured to discriminate whether the estimated target arterial waveform of the G1 output is true; A second discriminator D2 is configured to discriminate whether the reconstructed peripheral signal of the G2 output is true; s5.2.4 loss function definition and training the G1 and G2 networks to minimize a total loss function comprising at least a combination of one or more of: a. Physical loss The loss constraint of the residual error defined by the control equation which is not divided by the arterial modeling of the target blood pressure measurement section in the step 4, the blood pressure and blood flow waveform of the intermediate node of the G2 output must conform to the physical law of blood flow dynamics; b. reconstruction loss Quantifying the difference between the reconstructed peripheral signal output by the G2 and the reference peripheral signal; c. loss of resistance G1 and G2 are trained to deceive D1 and D2 so that the waveforms generated by the same are judged to be true by the discriminator; d. Loss of reference data During the training phase, the total loss function also includes a loss term for comparing the difference between the estimated brachial artery blood pressure waveform of the G1 output and the reference brachial artery blood pressure waveform.
5. 5. The method of claim 4, wherein the reconstructed peripheral signal of S5.2.2 is a radial artery end blood flow estimated from G2 integrated to obtain a reconstructed radial artery PPG signal.

Description

Multi-arterial coupling wearable continuous blood pressure measuring device and method based on multi-modal sensing Technical Field The invention relates to the technical field of medical monitoring, in particular to a blood pressure measuring method and device. Background Blood pressure is one of the most important vital sign parameters. Existing blood pressure measurement methods are mainly classified into direct measurement methods (i.e., invasive measurement) and indirect measurement methods (i.e., non-invasive measurement). Direct measurement measures the pressure in the artery directly through arterial puncture catheterization. The method has the advantages of accurate measurement result and continuous monitoring, and has the defects of traumatism, infection risk and operation of professional medical staff, so that the application scene is generally limited to critical care or large-scale operation. Indirect measurement is widely used because of its non-invasive nature. The most common indirect measurement is based on cuff pressurization, such as Korotkoff (Korotkoff) based auscultation and oscillography. However, the method has inherent technical defects that firstly, the cuff can cause compression and uncomfortable feeling to a tested person in the inflation and deflation process to disturb the normal physiological state, and secondly, the measurement is intermittent, continuous blood pressure reading cannot be provided, and real-time blood pressure fluctuation is difficult to capture. For continuous non-invasive measurements, methods such as radial artery applanation and arterial volume compensation have also been developed. While these methods can provide continuous readings, they still require a strong external pressure to the measurement site (e.g., applanation of the artery or application of back pressure), again causing discomfort. In addition, the measuring devices of these methods are generally complex in structure, bulky, and highly skilled in the art of operators, limiting their use in everyday and portable settings. Therefore, a blood pressure measurement technology which can realize noninvasive, undisturbed (i.e. no need of a cuff or strong external pressure) continuous and real-time is developed, and has great clinical significance for early screening, diagnosis and long-term health management of hypertension. With the development of wearable sensing technology and intelligent algorithms, there have been some advances in sleeveless blood pressure measurement methods, particularly methods based on photoplethysmogram (PPG) signals. Such methods typically use PPG and Electrocardiograph (ECG) signals to calculate Pulse TRANSIT TIME (PTT) and model between PTT and blood pressure. The measuring equipment has simple structure and good portability, and provides possibility for realizing the noninvasive, undisturbed and continuous blood pressure monitoring. In recent years, with rapid development of algorithms such as deep learning, a deep blood pressure estimation framework using PPG signals has been significantly advanced. However, the existing methods which rely heavily on PPG signals face two major limitations in practical application, so that the technical solution has inherent drawbacks, namely, one of them, the locality defect of the signals. PPG signals are typically acquired from a single location on the periphery (such as a finger or wrist) that reflects only local regional blood flow characteristics or blood volume changes. Such local signals are difficult to capture comprehensively the dynamics of the complex circulatory system of the human body. And secondly, the physiological distortion defect of the model. Most existing methods attempt to establish a direct mapping between peripheral PPG signal characteristics and brachial artery reference blood pressure. This direct mapping ignores the significant physiological differences in blood flow path (e.g., from the aorta to the brachial artery, to the radial artery) between different vessel segments, such as arterial wall elasticity, vessel stiffness, etc. The technical problem directly caused by the defects is that the existing PPG-based blood pressure estimation model has limited individuation capability and poor consistency of measurement results among different individuals. In order to solve the above-mentioned problem of cuff-free continuous blood pressure prediction, studies have been proposed to combine blood vessel information from arteries with PPG signals to improve estimation accuracy. In the field of hemodynamic studies, one-dimensional (1D) vessel models have been widely used to build personalized models based on patient specific anatomical and physiological data, while zero-dimensional (0D) models are also often used as boundary conditions for 1D models for coupling. The integration of the 0D-1D arterial model provides a theoretical framework for studying blood pressure propagation in the blood flow path, making it possible