Wearable multimodal hemodynamic monitoring device and method

Abstract

This invention relates to a wearable multimodal hemodynamic monitoring device and method, belonging to the field of biotechnology and medicine. The device uses fewer sensors to acquire a greater number of multimodal physiological signals at fewer points on the human body. After signal processing and parameter calculation, it can continuously and synchronously monitor hemodynamic parameters including blood flow, blood pressure, and vascular resistance. Compared with existing technologies, this invention has advantages such as comprehensive measurement parameters, portable device, long battery life, and good human-computer interaction. It can be used for long-term, real-time hemodynamic monitoring and has significant application value and broad application prospects in cardiovascular disease diagnosis, chronic disease management, and exercise assessment.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of biology and medicine, and relates to a wearable multi-modal hemodynamic monitoring device and method. BACKGROUND

[0002] Hemodynamics refers to the mechanics of blood flow in the cardiovascular system, and its basic research object is the relationship between blood flow, vascular resistance and blood pressure. Hemodynamic monitoring can obtain a large number of physiological parameters reflecting the function of the heart and the blood circulation system, such as heart rate (HR), stroke volume (SV), cardiac output (CO), blood pressure (BP), systemic vascular resistance (SVR), etc., which plays an important role in clinical vital sign monitoring and cardiovascular disease diagnosis and treatment.

[0003] In the past five decades of the development of hemodynamic monitoring, its application scenarios are usually hospital monitoring rooms or operating rooms, which is limited by the invasiveness of the monitoring method and the non-portability of the monitoring device.

[0004] With the improvement of people's pursuit of healthy life, compared with the treatment after the occurrence of disease, diagnosis and prevention before the deterioration of disease are particularly important, which can greatly reduce the pain of patients and medical expenses. Therefore, the control and management of cardiovascular diseases are crucial, and continuous and dynamic hemodynamic monitoring can provide support and help for this, and it is necessary to apply hemodynamic monitoring technology to a wider population, not just critically ill patients.

[0005] Although some existing portable or wearable hemodynamic commercial medical devices, such as ICON (capable of non-invasive continuous measurement of CO) developed by Osypka Medical Company in Germany, Enduro (capable of non-invasive continuous measurement of CO) developed by PhysioFlow Company in France, and ViSi (capable of non-invasive continuous measurement of BP) developed by Sotera Wireless Company in the United States, etc., its endurance cannot meet the monitoring needs of one day, and it does not have the ability of comprehensive hemodynamic monitoring (simultaneous acquisition of CO and BP parameters).

[0006] The existing detection method of wearable multi-modal physiological signals usually uses multiple non-invasive detection technologies and sensors to synchronously collect multiple signals reflecting different physiological states at multiple points on the human body surface, for example, multiple electrodes are used to detect electrocardiogram signals on the chest, multiple electrodes are used to detect thoracic impedance signals on the neck and chest, multiple photoelectric sensors are used to detect pulse wave signals on the fingers or wrists, piezoelectric sensors are used to detect heart sound signals on the chest, and acceleration sensors are used to detect heart vibration signals on the chest, etc.

[0007] Simultaneous monitoring of multimodal physiological signals presents numerous challenges to system design: from a physical perspective, it may increase the size and weight of the device; from a power consumption perspective, it may increase power consumption and reduce battery life; from a comfort perspective, it may reduce measurement comfort and restrict user activity; from a data processing perspective, it may increase algorithm complexity, multiply the amount of data, and affect the system's real-time performance.

[0008] Currently, wearable hemodynamic monitoring devices that combine advantages such as small size, light weight, low power consumption, long battery life, comprehensive hemodynamic parameters, and good ergonomics are still rare. Furthermore, after the sensor signals are acquired, they typically require preprocessing, feature extraction, and parameter calculation to obtain the measured parameters. Methods for extracting multimodal physiological signals from limited sensor signals are rarely reported. In addition, methods for combining different modal physiological signals to obtain multiple physiological parameters, especially comprehensive hemodynamic parameters including blood flow, blood pressure, and vascular resistance, have not yet been reported. Summary of the Invention

[0009] In view of this, the purpose of the present invention is to provide a wearable multimodal hemodynamic monitoring device and method to achieve continuous, comprehensive, long-term and real-time hemodynamic monitoring anytime and anywhere, and to broaden the application scenarios of hemodynamic monitoring.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A wearable multimodal hemodynamic monitoring device includes a main unit, a sensor box, and sensor cables; the sensor cables include a sensor bus and four lead wires; the sensor box includes an external expansion board; the four lead wires are connected to the sensor bus through the external expansion board;

[0012] The device host includes a motherboard, a battery, and a power board, a communication board, and an internal expansion board, all connected to the motherboard. The power board is connected to both the communication board and the battery. The motherboard is connected to an external expansion board via a sensor bus.

[0013] The motherboard includes a signal generation circuit, a signal detection circuit, a data processing circuit, a lead disconnection detection circuit, a defibrillation protection circuit, a power supply circuit, and peripheral circuits. The data processing circuit includes a microcontroller and its peripheral circuits for signal acquisition and processing. The microcontroller is connected to the signal generation circuit, signal detection circuit, lead disconnection detection circuit, power supply circuit, and peripheral circuits. The signal generation circuit and signal detection circuit acquire the patient's chest electrocardiogram (ECG) signals through four leads. The lead disconnection detection circuit determines the connection status of the four leads to the patient. The defibrillation protection circuit protects the device from damage during defibrillation.

[0014] Both the internal and external expansion boards include photoelectric sensors and photoelectric analog front-ends, triaxial accelerometers and temperature sensors, which are used to acquire photoelectric signals, triaxial acceleration signals and skin temperature at the distal and proximal ends of the heart, respectively; the photoelectric analog front-ends, triaxial accelerometers and temperature sensors of the internal and external expansion boards are all connected to the microcontroller.

[0015] Preferably, the photoelectric sensors all include multiple light-emitting diodes in the visible-near-infrared wavelength range and photodiodes in the corresponding wavelength range; the photoelectric signals are photoelectric signals of multiple wavelengths collected by the photoelectric sensors.

[0016] Preferably, the power supply circuit includes a power conversion circuit and a power control circuit; the power conversion circuit is used to generate different voltages required for the device to operate; the power control circuit is connected to a microcontroller and is used to control the device to turn on and off.

[0017] The peripheral circuitry includes a display driver circuit, a memory card peripheral circuit, a memory and its peripheral circuits, and a sound circuit; the memory and the sound circuit are respectively connected to the microcontroller and are used to store device information and generate prompt sounds.

[0018] Preferably, the power board includes an external interface, a power isolation circuit, a communication isolation circuit, a charge / discharge management circuit, a power monitoring circuit, and a battery protection circuit, which are respectively used for wired connection to external devices, power isolation, communication isolation, managing battery charge / discharge, acquiring battery information, and preventing abnormal battery electrical parameters.

[0019] Preferably, the communication board includes a wired transmission circuit and a wireless transmission circuit for establishing connections between the microcontroller and the wireless and wired terminals.

[0020] Preferably, the device host further includes a display screen, a memory card, and buttons; the buttons include a power button and a function button, which are respectively connected to the power control circuit and the microcontroller; the display screen, memory card, and function buttons are respectively connected to the microcontroller and are used to display information, save monitoring data, and trigger function settings.

[0021] Preferably, the main unit of the device is fixed to the wrist with a wristband, the sensor box is fixed to the chest with medical double-sided tape, the sensor bus runs from the lower arm through the upper arm to the chest, and the four lead wires are connected to four electrodes attached to the human chest.

[0022] Preferably, the main unit of the device is fixed to the waist with a belt, the sensor box is fixed to the chest with medical double-sided tape, the sensor bus is routed from the waist to the chest, and the four lead wires are connected to four electrodes attached to the human chest.

[0023] Furthermore, the monitoring methods of this device include multimodal physiological signal extraction methods, multimodal physiological signal processing methods, and multimodal hemodynamic parameter calculation methods;

[0024] The multimodal physiological signals consist of signals reflecting cardiac electrical characteristics, cardiac mechanical characteristics, vascular compliance, respiratory characteristics, blood circulation characteristics, and limb movement characteristics, including electrocardiogram signals, thoracic cavity baseline impedance signals, thoracic impedance change signals, thoracic impedance differential signals, respiratory signals (impedance), chest pulse signals, respiratory signals (photoelectric), chest movement signals, cardiac vibration signals, wrist pulse signals, wrist movement signals, chest temperature signals, and wrist temperature signals.

[0025] The multimodal physiological signal extraction method includes: extracting electrocardiogram signals, thoracic cavity baseline impedance signals, thoracic impedance change signals, thoracic impedance differential signals, and respiratory signals (impedance) from thoracic electrocardiogram signals; extracting chest pulse signals and respiratory signals (photoelectric signals) from proximal cardiac photoelectric signals; extracting chest motion signals and cardiac vibration signals from proximal cardiac triaxial acceleration signals; extracting wrist pulse signals from distal cardiac photoelectric signals; and extracting wrist (or lumbar) motion signals from distal cardiac triaxial acceleration signals.

[0026] The multimodal physiological signal processing method includes: denoising the multimodal physiological signals and locating the features of the multimodal physiological signals;

[0027] The method for calculating multimodal hemodynamic parameters includes: calculating cardiac output, blood pressure, and other hemodynamic parameters.

[0028] Furthermore, the method for extracting electrocardiogram signals, basic thoracic impedance signals, thoracic impedance change signals, thoracic impedance differential signals, and respiratory signals from chest electrical signals includes the following steps:

[0029] (1) Pass the chest electrocardiogram signal through a bandpass filter to obtain a low-frequency electrocardiogram signal;

[0030] (2) The chest electrical signal is filtered through a high-pass filter to obtain the chest impedance amplitude-modulated signal;

[0031] (3) Demodulate the amplitude-modulated chest impedance signal to obtain the chest impedance signal;

[0032] (4) The thoracic impedance signal is low-pass filtered to obtain the basic thoracic impedance signal;

[0033] (5) The chest impedance signal is filtered by a high-pass filter to obtain the chest impedance change signal;

[0034] (6) Baseline extraction of chest impedance signal to obtain respiratory signal (impedance);

[0035] (7) Take the first derivative (difference) of the chest impedance change signal to obtain the chest impedance differential signal;

[0036] The method for extracting chest pulse and respiratory signals from proximal cardiac photoelectric signals includes obtaining chest pulse signals through bandpass filtering and obtaining respiratory signals (photoelectric) through baseline extraction.

[0037] The method for extracting chest motion signals and cardiac vibration signals from proximal triaxial acceleration signals includes obtaining chest motion signals in the corresponding direction by low-pass filtering of the triaxial acceleration signals, and obtaining cardiac vibration signals by band-pass filtering of the acceleration signals perpendicular to the chest cavity.

[0038] The method for extracting wrist pulse signals from distal cardiac photoelectric signals includes obtaining wrist pulse signals through bandpass filtering;

[0039] The method for extracting wrist (or waist) motion signals from the distal triaxial acceleration signal of the heart includes obtaining wrist (or waist) motion signals in the corresponding direction by low-pass filtering the triaxial acceleration signal.

[0040] Furthermore, the method for denoising multimodal physiological signals includes the following steps:

[0041] (1) Preliminary denoising of multimodal physiological signals;

[0042] (2) Correcting the temporal relationship between multimodal physiological signals;

[0043] (3) Remove interference components from multimodal physiological signals;

[0044] (4) Select high-quality multimodal physiological signals;

[0045] The method for localizing multimodal physiological signal features includes the following steps:

[0046] (1) Obtain and process the first-order difference signal of the electrocardiogram signal;

[0047] (2) Determine the maximum value point of the first-order differential electrocardiogram signal in each cardiac cycle;

[0048] (3) Determine the Q point and R point of the electrocardiogram signal;

[0049] (4) Determine point C of the differential signal of the chest impedance;

[0050] (5) Determine point B of the differential signal of the chest impedance;

[0051] (6) Determine the O point of the differential signal of the chest impedance;

[0052] (7) Determine the X point of the differential signal of the chest impedance;

[0053] (8) Determine point C for the chest pulse signal and the wrist pulse signal respectively;

[0054] (9) Determine point B for the chest pulse signal and the wrist pulse signal respectively.

[0055] Furthermore, cardiac output was calculated using a stroke volume-thoracic impedance mathematical model; blood pressure was calculated using a multimodal data fusion blood pressure model; and other hemodynamic parameters were calculated using hemodynamic formulas related to cardiac output and blood pressure.

[0056] The beneficial effects of this invention are as follows:

[0057] (1) The device of the present invention adopts a multi-board design scheme, which has the advantages of small size, light weight, low power consumption, long battery life and good human-computer interaction. It can use fewer sensors to obtain more multimodal physiological signals at fewer points on the human body.

[0058] (2) The structure of the device of the present invention conforms to ergonomics and has two wearing methods: wrist wearing and waist wearing, which can adapt to the needs of different scenarios.

[0059] (3) The method of the present invention can extract multimodal physiological signals reflecting cardiac electrical characteristics, cardiac mechanical characteristics, vascular compliance, respiratory characteristics, blood circulation characteristics and limb movement characteristics from a variety of sensor signals, which is beneficial to the comprehensiveness of health assessment and the accuracy of parameter acquisition.

[0060] (4) The denoising method for multimodal physiological signals of the present invention uses a filter composed of causal systems, which has low computational load and good real-time performance, which is conducive to reducing hardware overhead. In addition, combining denoising and quality assessment of different modal signals is conducive to improving the signal-to-noise ratio and ensuring the effectiveness of the signal.

[0061] (5) This invention utilizes the temporal correlation of multimodal physiological signals and combines time-domain waveform features to locate feature points of electrocardiogram signals, chest impedance differential signals, chest pulse signals, and wrist pulse signals. The localization method has the characteristics of low complexity and high robustness.

[0062] (6) Based on multiple features extracted from multimodal physiological signals, the present invention uses mathematical models and multimodal data fusion models to calculate SV and BP respectively, and uses these to calculate other hemodynamic parameters, which has the advantage of comprehensive hemodynamic parameter monitoring including blood flow, vascular pressure and vascular resistance.

[0063] (7) The numerous advantages of the present invention described above make it of great application value and broad application prospects in cardiovascular disease diagnosis, chronic disease management and exercise assessment.

[0064] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0065] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0066] Figure 1 This is a structural diagram of the wearable multimodal hemodynamic monitoring device of the present invention;

[0067] Figure 2 This is a block diagram of the power conversion circuit in an embodiment of the present invention;

[0068] Figure 3 This is a block diagram of the power control circuit in an embodiment of the present invention;

[0069] Figure 4 This is a schematic diagram of the circuit formed by the power board in an embodiment of the present invention;

[0070] Figure 5 This is a connection diagram of the microcontroller and related components in an embodiment of the present invention;

[0071] Figure 6 This is a schematic diagram illustrating two wearing methods of the monitoring device in this embodiment of the invention: wrist wearing and waist wearing.

[0072] Figure 7 This is a block diagram of the wearable multimodal hemodynamic monitoring method of the present invention;

[0073] Figure 8 This is a flowchart illustrating the method for extracting electrocardiogram signals, basic thoracic impedance signals, thoracic impedance change signals, thoracic impedance differential signals, and respiratory signals (impedance) from thoracic electrical signals in an embodiment of the present invention.

[0074] Figure 9 This is a flowchart illustrating the methods for extracting chest pulse and respiratory signals (photoelectric) from proximal cardiac photoelectric signals and the method for extracting wrist pulse signals from distal cardiac photoelectric signals in embodiments of the present invention.

[0075] Figure 10 This is a flowchart illustrating the methods for extracting chest motion signals and cardiac vibration signals from proximal cardiac triaxial acceleration signals and the methods for extracting wrist (or waist) motion signals from distal cardiac triaxial acceleration signals in embodiments of the present invention.

[0076] Figure 11These are example waveforms of some multimodal physiological signals that are unaffected and affected by interference in embodiments of the present invention;

[0077] Figure 12 The following are waveform examples of chest Z-axis motion signals and cardiac vibration signals extracted from two segments of proximal cardiac Z-axis acceleration signals under different motion states in an embodiment of the present invention.

[0078] Figure 13 This is a flowchart of the multimodal physiological signal denoising method of the present invention;

[0079] Figure 14 This is a flowchart of an algorithm for removing respiratory interference components from chest pulse signals using an adaptive filter in an embodiment of the present invention.

[0080] Figure 15 This is an example diagram illustrating the effect of using an adaptive filter to remove respiratory interference components from chest pulse signals in an embodiment of the present invention.

[0081] Figure 16 This is a flowchart of the multimodal physiological signal feature localization method of the present invention;

[0082] Figure 17 This is an example diagram illustrating the effect of the multimodal physiological signal feature localization method in an embodiment of the present invention;

[0083] Figure 18 This is a schematic diagram illustrating the calculation of multimodal hemodynamic parameters in an embodiment of the present invention. Detailed Implementation

[0084] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0085] Please see Figures 1-18 This invention provides a wearable multimodal hemodynamic monitoring device, such as... Figure 1 As shown, the device includes a main unit, a sensor box, and sensor cables. The sensor cables include a sensor bus and four lead wires. The four lead wires connect to the sensor box; the four lead wires and the sensor box are connected to the main unit via the sensor bus.

[0086] The main unit of the device consists of several components, including a motherboard, power board, communication board, internal expansion board, display screen, memory card, buttons, and battery. The display screen and buttons are used for human-machine interaction, displaying information and controlling the device, respectively. The buttons consist of function keys and a power button, used for function settings and turning the device on and off, respectively. The memory card is used to store monitoring data. The battery provides power for the device to operate.

[0087] The motherboard includes a signal generation circuit, a signal detection circuit, a data processing circuit, a lead disconnection detection circuit, a defibrillation protection circuit, a power supply circuit, and peripheral circuits. These circuits are used to generate current excitation signals, detect chest electrical signals, process signals through a microcontroller and control the device's operating status, detect lead connection status, protect components from defibrillation damage, perform voltage conversion and power control, and drive components other than the microcontroller.

[0088] Optionally, for an organization and implementation scheme of the signal generation circuit, signal detection circuit and microcontroller, please refer to the patent application with publication number CN117617937A, entitled Hemodynamic Detection System and Method Based on Fewer Electrodes and Signal Correlation.

[0089] Optionally, the signal detection circuit includes an optoelectronic signal conditioning circuit.

[0090] Optionally, for an organization and implementation scheme of the lead detachment detection circuit and the defibrillation protection circuit, please refer to the patent application with publication number CN118557170A, entitled "A Method and Implementation Circuit for Identifying Lead Detachment in Chest Electrophysiological Impedance Detection".

[0091] Optionally, the current lead disconnection detection is performed by detecting the current in the current lead circuit.

[0092] Optionally, the power supply circuit includes a power conversion circuit and a power control circuit, which are used to convert voltage to generate different voltages required for the device to operate, and to control the device to turn on and off.

[0093] The composition of the power conversion circuit is as follows: Figure 2 As shown, VSYS generates 3.3V, 2.5V, and -2.5V sequentially through a voltage conversion circuit to power digital and analog devices.

[0094] The power control circuit controls the presence or absence of the system power supply VSYS based on the microcontroller signal VSYS_IO and the on / off switch signal KEY_IO, thereby controlling the device's on / off state. Figure 3As shown. The self-locking circuit functions as follows: after power-on (when VIN has voltage), IN defaults to a high level, and OUT defaults to a low level; when IN transitions from a high level to a low level and remains low for several seconds, OUT outputs the opposite level. The electronic switch circuit functions as follows: when OUT is high, the switch is closed, connecting VIN and VSYS, thus powering the device; conversely, when OUT is low, the switch is open, and the VSYS voltage is 0. The pull-up resistor at the AND gate input has a resistance of 10kΩ to 100kΩ.

[0095] Upon power-up, VSYS_IO and KEY_IO are both high, IN is high, and OUT is low, the switch is open, and VSYS voltage is 0. When the power button is pressed, IN outputs a low level; after a few seconds, OUT goes high, the switch closes, and VSYS voltage is approximately equal to VIN, turning the device on. When the power button is pressed again for a few seconds, the switch opens, and the device shuts down. When the device is on, a sustained low level in VSYS_IO for several seconds will also shut it down. Therefore, the power control circuit can achieve the functions of turning the device on or off by pressing and holding the power button, as well as microcontroller-controlled shutdown.

[0096] Optionally, the power button is also connected to the microcontroller and can function as a function key when pressed briefly.

[0097] The power board includes an external interface, power isolation circuit, communication isolation circuit, charge / discharge management circuit, power monitoring circuit, and battery protection circuit. These are used for wired connection to external devices, power isolation, communication isolation, managing battery charge / discharge, acquiring battery information, and preventing abnormal battery electrical parameters, respectively. A schematic diagram of the circuit is shown below. Figure 4 As shown.

[0098] Optionally, the external interface uses a USB interface, and the external power supply is V. s The voltage is 5V, and the communication lines are differential lines D+ and D-. The battery is a lithium polymer battery with a nominal voltage of 3.7V. When connecting to an external interface, V s The isolation terminal power supply voltage V is generated by the power isolation circuit. si V si ≈V s The charge / discharge management circuit output voltage VIN automatically switches from the battery voltage to V. si At the same time, through V si Charge the battery. The battery protection circuit has functions such as overcharge, over-discharge, overcurrent, and short circuit protection. The power monitoring circuit can detect the battery voltage and sample the current through the sampling resistor R. s The charging current is detected to estimate the battery's remaining power (SOC) and other information. Communication isolation circuitry isolates the communication lines.

[0099] The communication board includes wired and wireless transmission circuits, used to establish connections between the microcontroller and wired and wireless terminals. Wireless transmission can be achieved via Bluetooth or WiFi.

[0100] The sensor box includes an external expansion board. Both the external and internal expansion boards consist of a photoelectric sensor, a photoelectric analog front-end, a triaxial accelerometer, and a temperature sensor, used to acquire photoelectric signals from the proximal and distal ends of the heart, triaxial acceleration signals, and skin temperature, respectively. The photoelectric signals are multi-wavelength photoelectric signals collected by the photoelectric sensor.

[0101] Please see Figure 5 Microcontrollers use multiple internal resources to establish connections with external circuits or sensors.

[0102] One way to wear monitoring equipment is as follows: Figure 6 As shown in (a), the main unit of the device is fixed to the wrist with a wristband, and the sensor box is fixed to the chest with medical double-sided tape. The sensor bus runs from the lower arm through the upper arm to the chest, and the four lead wires are connected to four electrodes attached to the chest. This wrist-worn method allows for the acquisition of physiological signals from the wrist, suitable for comprehensive monitoring in a resting state. Another wearing method for the monitoring device is as follows... Figure 6 As shown in (b), the main unit of the device is secured to the waist with a belt, and the sensor box is secured to the chest with medical double-sided tape. The sensor bus runs from the waist to the chest, and four leads connect to four electrodes attached to the chest. This waist-wearing method is more convenient and suitable for dynamic monitoring. The four electrodes are positioned as follows: the two middle electrodes (the second and third electrodes) are placed at the base of the neck on the left side of the body and at the level of the xiphoid process on the chest, respectively; the top electrode (the first electrode) is placed 3 cm above the second electrode; and the bottom electrode (the fourth electrode) is placed 5 cm below the third electrode.

[0103] The monitoring method for the wearable multimodal hemodynamic monitoring device provided by this invention includes three aspects: a multimodal physiological signal extraction method, a multimodal physiological signal processing method, and a multimodal hemodynamic parameter calculation method, as detailed below. Figure 7 As shown. Among them, the multimodal physiological signals consist of signals reflecting cardiac electrical characteristics, cardiac mechanical characteristics, vascular compliance, respiratory characteristics, blood circulation characteristics, and limb movement characteristics, including electrocardiogram signals, thoracic cavity baseline impedance signals, thoracic impedance change signals, thoracic impedance differential signals, respiratory signals (impedance), chest pulse signals, respiratory signals (photoelectric), chest movement signals, cardiac vibration signals, wrist pulse signals, wrist movement signals, chest temperature signals, and wrist temperature signals.

[0104] The purpose of multimodal physiological signal extraction methods is to extract multimodal physiological signals from multiple sensor signals, including: methods for extracting electrocardiogram signals, basic thoracic impedance signals, thoracic impedance change signals, thoracic impedance differential signals, and respiratory signals (impedance) from thoracic electrocardiogram signals; methods for extracting chest pulse signals and respiratory signals (photoelectric signals) from proximal cardiac photoelectric signals; methods for extracting chest motion signals and cardiac vibration signals from proximal cardiac triaxial acceleration signals; methods for extracting wrist pulse signals from distal cardiac photoelectric signals; and methods for extracting wrist (or lumbar) motion signals from distal cardiac triaxial acceleration signals.

[0105] Please see Figure 8 The chest electrocardiogram (ECG) signal is processed through high-pass and band-pass filtering to obtain a high-frequency chest impedance amplitude-modulated signal and a low-frequency ECG signal. The low-frequency envelope of the chest impedance amplitude-modulated signal is the chest impedance signal, which can be obtained through demodulation. Further high-pass, low-pass filtering, and baseline extraction are performed on the chest impedance signal to obtain the chest impedance change signal, the baseline chest impedance, and the respiratory signal (impedance). Taking the first derivative (difference) of the chest impedance change signal yields the chest impedance change differential signal, also known as the chest impedance differential signal.

[0106] Because the proximal cardiac photoelectric sensor is placed in the chest area, the photoelectric signal contains not only components caused by vascular pulsation but also low-frequency components caused by respiration. See also... Figure 9 The proximal cardiac photoelectric signals were bandpass filtered and baseline extracted to obtain chest pulse and respiratory signals (photoelectric). The distal cardiac photoelectric signals were bandpass filtered to obtain wrist pulse signals.

[0107] Please see Figure 10 The triaxial acceleration signals from the proximal and distal ends of the heart reflect the motion information of the chest and wrist. By extracting the baseline from the triaxial acceleration signals, the motion signals of the corresponding parts on the three axes can be obtained. Since the heartbeat causes displacement changes on the surface of the thoracic cavity, the oscillation signal can be extracted from the Z-axis acceleration signal from the proximal end of the heart.

[0108] The monitoring equipment acquires some multimodal physiological signals, which are then synchronously displayed on a host computer. Example waveforms showing waveforms unaffected and affected by respiration are shown below. Figure 11 As shown in the figure. Among them, ECG is the electrocardiogram signal, ICG is the differential signal of chest impedance, ΔZ is the signal of chest impedance change, cPPG is the chest pulse signal, wPPG is the wrist pulse signal, Z0&RESP is the mixed signal of chest cavity baseline impedance and respiratory signal (impedance), and cx, cy, and cz are the acceleration signals of the chest X-axis, Y-axis, and Z-axis, respectively.

[0109] Taking the proximal cardiac Z-axis acceleration signal as an example, a clear cardiac vibration signal can be extracted from it in a resting state; during the process from a resting state to walking forward, the baseline of the signal reflects the motion of the chest along the Z-axis, such as... Figure 12 As shown.

[0110] Please see Figure 13 The multimodal physiological signal denoising process involves four steps: first, preliminary denoising is achieved using a passband linear phase filter; second, an all-pass filter is used to compensate for the different time delays generated during signal processing in different channels, ensuring the correct timing relationship between signals; then, an adaptive filter is used to remove interference components from the signal; and finally, signal quality assessment is performed to obtain a multimodal physiological signal with better quality.

[0111] Taking the chest pulse signal as an example, firstly, low-pass and high-pass Bessel filters are used to remove high-frequency noise and suppress baseline drift, respectively; secondly, an all-pass filter is used to compensate for the time difference between it and the signal with the maximum group delay; then, the respiratory signal (impedance) is used as a reference signal, and an adaptive filter is used to remove respiratory interference; finally, based on the characteristics of the signal and combined with the chest motion signal in the same time period, the signal quality is evaluated, and the signal with better quality is selected.

[0112] Since the respiratory interference component in the chest pulse signal is highly correlated with the respiratory signal (impedance), an adaptive filter is used to remove the respiratory interference. Please refer to the algorithm block diagram. Figure 14 The adaptive algorithm employs the least mean square (LMS) algorithm, continuously adjusting the filter's weight parameters based on the noise reference signal and the error signal to minimize the mean square value of the error signal. An FIR filter is used as the adjustable filter, and its output is the convolution of the noise reference signal and the weight parameters. The iterative formula for the weight parameters is as follows:

[0113] w(n+1)=w(n)+2μe(n)x(n)

[0114] Where μ is the step size factor, and e(n) and x(n) are the error signal and the noise reference signal, respectively.

[0115] Please see Figure 15 The respiratory interference component of the chest pulse signal can be effectively removed using an adaptive filter.

[0116] Optionally, when the monitoring device is worn on the waist, the acceleration signals have a strong correlation because both the proximal and distal triaxial accelerometers of the heart are parallel to the frontal plane and located in the upper body. By using the distal Z-axis acceleration signal as a noise reference signal and employing an adaptive filter to remove motion interference from the proximal Z-axis acceleration signal, a high-quality cardiac vibration signal can be obtained during exercise.

[0117] In the signal quality assessment process, the initial quality assessment is first performed based on the energy characteristics of the non-motion signals from the chest or wrist. The segments with poor quality are denoted as P. Then, the motion segments Q are extracted based on the characteristics of the motion signals from the corresponding parts. Finally, the segments with poor quality are assessed as P∪Q, and the segments with good quality are Ω-P∪Q, where Ω represents the entire signal segment.

[0118] Optionally, the preliminary quality assessment can refer to the signal quality assessment method proposed in the patent application with publication number CN116451110A (invention title: Method for constructing a blood glucose prediction model based on signal energy characteristics and pulse cycle).

[0119] Multimodal physiological signal feature localization methods utilize the temporal relationships between different modal signals. (See also: [link to relevant documentation]). Figure 16 Specifically, it includes the following steps:

[0120] (1) Obtain the first-order difference signal of the electrocardiogram signal, square the elements greater than 0, set the remaining elements to 0, and use the relative energy method to extract the pulse to obtain the processed first-order difference signal of the electrocardiogram.

[0121] (2) Using the maximum value of the processed first-order differential ECG signal as a reference, the maximum value of the first-order differential ECG signal in each cardiac cycle is determined by the threshold method, and the interval between the maximum values ​​of adjacent cardiac cycles is used as the RR interval.

[0122] (3) Based on the maximum value of the first-order differential signal of the ECG in each cardiac cycle, the minimum value within a range of 0.1 seconds forward in the ECG signal is the ECG_Q point, and the maximum value within a range of 0.1 seconds backward is the ECG_R point.

[0123] (4) Using the ECG_R point as the reference, the maximum value point within one-third of the RR interval in the differential signal of chest impedance is the ICG_C point;

[0124] (5) The minimum point within the interval formed by the ECG_Q point and the ECG_R point in the differential signal of chest impedance is the ICG_B0 point.

[0125] (6) Use the least squares method of two line segments to determine the ICG_B point in the interval formed by the ICG_B0 point and the ICG_C point of the differential signal of the chest impedance;

[0126] (7) Search for the minimum point within one-third of the RR interval after the ICG_C point of the chest impedance differential signal, which is the ICG_X0 point;

[0127] (8) Using ICG_C as a fixed point, the shortest distance method is used to determine the ICG_O point within one-third of the RR interval after the ICG_X0 point of the chest impedance differential signal;

[0128] (9) The maximum slope method is used to determine the ICG_X point in the first three-quarters of the interval formed by the ICG_X0 point and the ICG_O point of the chest impedance differential signal.

[0129] (10) Using the ICG_C point of the chest impedance differential signal as a reference, search for the maximum value point within one-third of the RR interval in the chest pulse signal and wrist pulse signal respectively as the cPPG_C point and wPPG_C point;

[0130] (11) Search for the minimum points cPPG_B0 and wPPG_B0 within one-third of the RR intervals before the cPPG_C and wPPG_C points in the chest pulse signal and wrist pulse signal, respectively.

[0131] (12) Search for the minimum points cPPG_B0 and wPPG_B0 within the first third of the RR interval of the cPPG_C and wPPG_C points in the chest pulse signal and wrist pulse signal, respectively.

[0132] The cPPG_B point is determined within the interval formed by cPPG_B0 and cPPG_C points of the chest pulse signal using the bisegment least squares method, and the wPPG_B point is determined within the interval formed by wPPG_B0 and wPPG_C points of the wrist pulse signal.

[0133] The relative energy method is helpful in extracting the pulse component of a signal, and this method is widely used in biomedical signal processing. The coefficient signal c(n) is defined as the ratio of the short-term energy signal s(n) to the long-term energy signal l(n) of the original signal x(n):

[0134]

[0135] Among them, s w and l w represents the half-lengths of the short-term and long-term sliding windows, respectively, and p and w represent the exponent and window function of interest, respectively. In this embodiment, p is set to 2, and w uses a Hamming window. The output signal is the product of the coefficient signal and the original signal:

[0136] y(n) = c(n) × x(n)

[0137] The notch before point ICG_C is related to point ICG_B. Point ICG_B can be located well using the bisegment least squares method, which is defined as follows:

[0138]

[0139] Here, minError(x,l,r) represents the point obtained using the bisegment least squares method based on the sequence x and the search interval (l,r).

[0140] The goal of this optimization condition is to find a point between two fixed points l and r in sequence x such that this point forms a sequence with the linear sequences determined by the two fixed points. The error with the original sequence x is minimized.

[0141] The ICG_X point is closely related to the sharp rise in the differential signal of the posterior chest impedance at the end of the T wave in the electrocardiogram (ECG) signal. Therefore, the maximum slope method can be used to locate the X point, which is expressed by the following formula:

[0142]

[0143] The optimization condition represented by maxSlope(x,m,l,r) aims to find the point with the largest slope of the line segment formed by the fixed point m within the search interval (l,r).

[0144] Locating the ICG_X point using this method requires using the ICG_O point as a fixed point. The ICG_O point is typically defined as the maximum peak point within the CC interval. However, multiple peaks may exist within the CC interval, and the ICG_O point is not necessarily the maximum value point. To improve the accuracy of ICG_O point positioning, the ICG_O point is determined by the minimum distance between a point within the specified interval and the ICG_C point at a certain numerical ratio, specifically expressed by the following formula:

[0145]

[0146] Where minDistance(x,m,l,r,k) represents the point obtained using the shortest distance method based on the sequence x, the fixed point m, the search interval (l,r), and the scaling factor k; k can be determined by the ratio of the difference between the peak and trough values ​​of the signal to the number of points in the δ-second signal:

[0147]

[0148] Among them, f s This indicates the sampling frequency. In this embodiment, δ is set to 0.6 to locate the ICG_O point.

[0149] Please see Figure 17The markers for each cardiac cycle in the ECG signal, from left to right, are ECG_Q, ECG_R, and ECG_S. The marker for each cardiac cycle in the processed first-order differential ECG signal is the maximum value of the first-order differential ECG signal. The markers for each cardiac cycle in the chest impedance differential signal, from left to right, are ICG_B0, ICG_B, ICG_C, ICG_X, and ICG_O. The marker for each cardiac cycle in the chest impedance change signal is ΔZ_B (consistent with the horizontal position of ICG_B). The markers for each cardiac cycle in the chest pulse signal, from left to right, are cPPG_B0, cPPG_B, and cPPG_C. The markers for each cardiac cycle in the wrist pulse signal, from left to right, are wPPG_B0, wPPG_B, and wPPG_C. The multimodal physiological signal feature localization method can accurately locate signal feature points.

[0150] Multimodal hemodynamic parameter calculation methods can calculate multimodal hemodynamic parameters such as cardiac output and blood pressure based on signal characteristics.

[0151] The method for calculating cardiac output is based on a mathematical model of stroke volume versus chest impedance. The model uses a modified Sramek-Bernstein formula:

[0152]

[0153] Where SV is the stroke volume, |dZ / dt| max denoted as C, where C is the amplitude of the differential signal of the thoracic impedance; Z0 is the average value of the baseline thoracic impedance signal; LVET is the left ventricular ejection time; H is the height; δ is the ratio of actual weight to ideal weight; and α is the correction factor related to the equipment.

[0154] The blood pressure calculation method is based on a multimodal data fusion blood pressure model. The model's input variables consist of multiple features, including all variables in the modified Sramek-Bernstein formula, heart rate, and time variables such as ejection time (PEP) and chest pulse wave conduction time (cPTT), determined by the ECG_Q point, the differential chest impedance signal (ICG_B point), the chest pulse wave signal (cPPG_B point), and the wrist pulse wave signal (wPPG_B point). The model's intermediate layers consist of multiple artificial neural networks. The model's output variables include systolic and diastolic blood pressure.

[0155] Optionally, the input variables for the multimodal data fusion blood pressure model may also include wrist pulse wave arrival time (wPTT).

[0156] Other methods for calculating hemodynamic parameters are based on hemodynamic formulas related to cardiac output and blood pressure.

[0157] Please see Figure 18Based on signal characteristics, combined with the stroke volume-thoracic impedance mathematical model, multimodal data fusion blood pressure model, and relevant hemodynamic formulas, multimodal hemodynamic parameters such as cardiac output, mean blood pressure, and systemic vascular resistance can be calculated to comprehensively assess physiological states such as blood flow, blood pressure, vascular resistance, myocardial contractility, and fluid levels.

[0158] Optionally, systolic and diastolic blood pressure can be obtained via external input.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A wearable multimodal hemodynamic monitoring device, characterized in that, The device includes a main unit, a sensor box, and sensor cables; the sensor cables include a sensor bus and four lead wires; the sensor box includes an external expansion board; the four lead wires are connected to the sensor bus through the external expansion board. The device host includes a motherboard, a battery, and a power board, a communication board, and an internal expansion board, all connected to the motherboard. The power board is connected to the communication board and the battery. The motherboard is connected to the external expansion board via a sensor bus. The motherboard includes a signal generation circuit, a signal detection circuit, a data processing circuit, a lead disconnection detection circuit, a defibrillation protection circuit, a power supply circuit, and peripheral circuits. The data processing circuit includes a microcontroller and its peripheral circuits for signal acquisition and processing. The microcontroller is connected to the signal generation circuit, signal detection circuit, lead disconnection detection circuit, power supply circuit, and peripheral circuits. The signal generation circuit and signal detection circuit acquire the patient's chest electrocardiogram (ECG) signals through four leads. The lead disconnection detection circuit determines the connection status of the four leads to the patient. The defibrillation protection circuit protects the device from damage during defibrillation. Both the internal and external expansion boards include photoelectric sensors and photoelectric analog front-ends, triaxial accelerometers and temperature sensors, which are used to acquire photoelectric signals, triaxial acceleration signals and skin temperature at the distal and proximal ends of the heart, respectively; the photoelectric analog front-ends, triaxial accelerometers and temperature sensors of the internal and external expansion boards are all connected to the microcontroller; The monitoring methods of this device include multimodal physiological signal extraction methods, multimodal physiological signal processing methods, and multimodal hemodynamic parameter calculation methods; The multimodal physiological signals consist of signals reflecting cardiac electrical characteristics, cardiac mechanical characteristics, vascular compliance, respiratory characteristics, blood circulation characteristics, and limb movement characteristics, including electrocardiogram signals, thoracic cavity baseline impedance signals, thoracic impedance change signals, thoracic impedance differential signals, respiratory signals, chest pulse signals, respiratory signals, chest movement signals, cardiac vibration signals, wrist pulse signals, wrist movement signals, chest temperature signals, and wrist temperature signals. The multimodal physiological signal extraction method includes: extracting electrocardiogram signals, thoracic cavity baseline impedance signals, thoracic impedance change signals, thoracic impedance differential signals, and respiratory signals from chest electrocardiogram signals; extracting chest pulse signals and respiratory signals from proximal cardiac photoelectric signals; extracting chest motion signals and cardiac vibration signals from proximal cardiac triaxial acceleration signals; extracting wrist pulse signals from distal cardiac photoelectric signals; and extracting wrist or waist motion signals from distal cardiac triaxial acceleration signals. The multimodal physiological signal processing method includes: denoising the multimodal physiological signals and locating the features of the multimodal physiological signals; locating the features of the multimodal physiological signals using the temporal relationship between different modal signals specifically includes the following steps: (1) Obtain the first-order difference signal of the electrocardiogram signal, square the elements greater than 0, set the rest of the elements to 0, and use the relative energy method to extract the pulse to obtain the processed first-order difference signal of the electrocardiogram. (2) Using the maximum value of the processed first-order differential ECG signal as a reference, the maximum value of the first-order differential ECG signal in each cardiac cycle is determined by the threshold method, and the interval between the maximum values ​​of adjacent cardiac cycles is used as the RR interval. (3) Based on the maximum value of the first-order differential signal of the ECG in each cardiac cycle, search for the minimum value within a range of 0.1 seconds forward in the ECG signal as the ECG_Q point, and search for the maximum value within a range of 0.1 seconds backward as the ECG_R point; (4) Using the ECG_R point as a reference, the maximum value point within one-third of the RR interval in the differential signal of chest impedance is the ICG_C point; (5) The minimum point within the interval formed by the ECG_Q point and the ECG_R point in the differential signal of chest impedance is the ICG_B0 point; (6) Use the two-segment least squares method to determine the ICG_B point in the interval formed by the ICG_B0 point and the ICG_C point of the chest impedance differential signal; (7) The minimum point is found within one-third of the RR interval after the ICG_C point of the chest impedance differential signal; (8) Using ICG_C as a fixed point, determine ICG_O within one-third of the RR interval after ICG_X0 of the chest impedance differential signal using the shortest distance method; (9) Use the maximum slope method to determine the ICG_X point in the first three-quarters of the interval formed by the ICG_X0 point and the ICG_O point of the chest impedance differential signal; (10) Using the ICG_C point of the differential signal of chest impedance as a reference, search for the maximum value point within one-third of the RR interval in the chest pulse signal and wrist pulse signal respectively as the cPPG_C point and wPPG_C point; (11) Search for the minimum value points cPPG_B0 and wPPG_B0 points respectively within the first third of the RR interval of the cPPG_C point and wPPG_C point in the chest pulse signal and wrist pulse signal; (12) Search for the minimum value points cPPG_B0 and wPPG_B0 points respectively within the first third of the RR interval of the cPPG_C point and wPPG_C point in the chest pulse signal and wrist pulse signal; The cPPG_B point is determined within the interval formed by cPPG_B0 and cPPG_C points of the chest pulse signal using the double-segment least squares method, and the wPPG_B point is determined within the interval formed by wPPG_B0 and wPPG_C points of the wrist pulse signal. The method for calculating multimodal hemodynamic parameters includes: Cardiac output was calculated using a stroke volume-thoracic impedance mathematical model; this model employed a modified Sramek-Bernstein formula. Where SV is the stroke volume, | dZ / dt | max The amplitude of the differential signal of the thoracic impedance at point C, i.e., ICG_C, is the basic impedance of the thoracic cavity. Z 0 represents the average value of the baseline thoracic impedance signal, and LVET represents the left ventricular ejection time. H For height, δ It is the ratio of actual weight to ideal weight. α These are correction factors related to the equipment. Blood pressure is calculated using a multimodal data fusion blood pressure model. The input variables of the multimodal data fusion blood pressure model consist of multiple features, including all variables in the modified Sramek-Bernstein formula, heart rate, and time variables such as ejection time (PEP) and chest pulse wave conduction time (cPTT) determined by the ECG_Q point, chest impedance differential signal (ICG_B point), chest pulse wave signal (cPPG_B point), and wrist pulse wave signal (wPPG_B point). The intermediate layer of the model consists of multiple artificial neural networks. The output variables of the model include systolic blood pressure and diastolic blood pressure. Hemodynamic parameters other than cardiac output and blood pressure are calculated using hemodynamic formulas related to cardiac output and blood pressure.

2. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, The power supply circuit includes a power conversion circuit and a power control circuit; the power conversion circuit is used to generate different voltages required for the device to operate; the power control circuit is connected to a microcontroller and is used to control the device to turn on and off. The peripheral circuitry includes a display driver circuit, a memory card peripheral circuit, a memory and its peripheral circuits, and a sound circuit; the memory and the sound circuit are respectively connected to the microcontroller and are used to store device information and generate prompt sounds.

3. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, The power board includes an external interface, a power isolation circuit, a communication isolation circuit, a charge / discharge management circuit, a power monitoring circuit, and a battery protection circuit, which are used for wired connection to external devices, power isolation, communication isolation, management of battery charge / discharge, acquisition of battery information, and prevention of abnormal battery electrical parameters, respectively.

4. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, The device host also includes a display screen, a memory card, and buttons; the buttons include a power button and a function button, which are respectively connected to the power control circuit and the microcontroller; the display screen, memory card, and function buttons are respectively connected to the microcontroller and are used to display information, save monitoring data, and trigger function settings.

5. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, The main unit of the device is fixed to the wrist, the sensor box is fixed to the chest, the sensor bus runs from the lower arm through the upper arm to the chest, and the four lead wires are connected to four electrodes attached to the human chest.

6. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, The main unit of the device is fixed to the waist, the sensor box is fixed to the chest, the sensor bus is routed from the waist to the chest, and the four lead wires are connected to four electrodes attached to the chest.

7. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, The method for extracting electrocardiogram signals, basic thoracic impedance signals, thoracic impedance change signals, thoracic impedance differential signals, and respiratory signals from chest electrical signals includes the following steps: (1) Pass the chest electrocardiogram signal through a bandpass filter to obtain a low-frequency electrocardiogram signal; (2) The chest electrical signal is passed through a high-pass filter to obtain the chest impedance amplitude-modulated signal; (3) Demodulate the amplitude-modulated chest impedance signal to obtain the chest impedance signal; (4) The thoracic impedance signal is low-pass filtered to obtain the basic thoracic impedance signal; (5) The chest impedance signal is high-pass filtered to obtain the chest impedance change signal; (6) Baseline extraction of chest impedance signal to obtain respiratory signal; (7) Take the first derivative of the chest impedance change signal to obtain the chest impedance differential signal; The method for extracting chest pulse and respiratory signals from proximal cardiac photoelectric signals includes obtaining chest pulse signals through bandpass filtering and obtaining respiratory signals through baseline extraction. The method for extracting chest motion signals and cardiac vibration signals from proximal triaxial acceleration signals includes obtaining chest motion signals in the corresponding direction by low-pass filtering of the triaxial acceleration signals, and obtaining cardiac vibration signals by band-pass filtering of the acceleration signals perpendicular to the chest cavity. The method for extracting wrist pulse signals from distal cardiac photoelectric signals includes obtaining wrist pulse signals through bandpass filtering; The method for extracting wrist or waist motion signals from the distal triaxial acceleration signal of the heart includes obtaining wrist or waist motion signals in the corresponding direction by low-pass filtering the triaxial acceleration signal.

8. The wearable multimodal hemodynamic monitoring device according to claim 1, characterized in that, Methods for denoising multimodal physiological signals include the following steps: (1) Preliminary denoising of multimodal physiological signals; (2) Correcting the temporal relationships between multimodal physiological signals; (3) Remove interference components from multimodal physiological signals; (4) Select high-quality multimodal physiological signals.

Images