A dynamic non-invasive cardiac output monitoring system and method

Abstract

The present invention belongs to the technical field of non-invasive hemodynamic monitoring, and provides a dynamic non-invasive cardiac output monitoring system and method. Based on a sensor array, high-quality dynamic pulse wave waveform signals are collected and motion artifacts in the motion state are corrected to meet the input conditions of the cardiac output calculation model, and dynamic pulse wave feature vectors are calculated; an estimated value of the aortic cross-sectional area is calculated as a correction variable based on personalized characteristic parameters; based on a cardiac output model trained by deep learning, the cardiac output is calculated to realize real-time monitoring and evaluation of the dynamic cardiac output of the subject. The present invention can realize dynamic, real-time, and long-term monitoring of cardiac output parameters, and does not require intermittent correction, and is suitable for home rehabilitation use by patients with heart diseases.

Description

Technical Field

[0001] This application relates to the technical field of hemodynamic monitoring, and in particular, to a dynamic non-invasive cardiac output monitoring system and method. Background Art

[0002] Non-invasive and long-term cardiac output parameter monitoring is of great significance for home rehabilitation function monitoring of patients with heart failure and structural heart diseases. However, the existing technologies and products have not yet truly achieved long-term and dynamic monitoring of cardiac output. There are patents that can establish a model related to cardiac output through pulse wave characteristics and patients' personalized information. Among them, some technologies require using a sphygmomanometer to correct the pulse wave characteristics of peripheral arteries at certain time intervals, which is not suitable for the requirements of long-term monitoring scenarios. Patent CN103784132A can measure cardiac output under resting conditions, but does not address the extraction method of feature vectors under active or exercise conditions; Patent CN102499669A can perform waveform profile analysis on the pulse wave waveform under exercise conditions to obtain the pulse wave arrival time and the ratio of the pulse wave inflection point area as feature vectors under exercise conditions, but fails to accurately calculate cardiac output. Therefore, it is necessary to develop a high-quality dynamic cardiac output monitoring model to meet the needs of long-term monitoring of home rehabilitation of cardiovascular patients. Summary of the Invention

[0003] In order to solve the technical problem that the prior art cannot achieve dynamic, continuous, and long-term high-precision monitoring of cardiac output, the present invention provides a dynamic non-invasive cardiac output monitoring system and method, which can not only directly collect multi-channel photoplethysmogram signals and accelerometer signals at different peripheral artery sites under resting or active states and filter and amplify them to obtain dynamic pulse wave waveform signals, but also be applied to resting states and daily exercise scenarios, realizing dynamic measurement of individual cardiac output, which is of great significance for home rehabilitation monitoring of patients with heart failure and structural heart diseases.

[0004] The specific solutions are as follows:

[0005] A dynamic non-invasive cardiac output monitoring system includes: a signal acquisition module, a filtering module, a pulse wave feature extraction module, a dynamic signal quality monitoring module, and a cardiac output monitoring and evaluation module;

[0006] The signal acquisition module is used to continuously collect multi-channel photoplethysmogram signals and accelerometer signals at different peripheral artery sites under resting or active states, filter and amplify them to obtain dynamic pulse wave waveform signals;

[0007] The filtering module can correct baseline drift and motion artifacts in the dynamic pulse wave waveform signals by combining accelerometer signals and an adaptive filtering algorithm to obtain pulse wave corrected signals;

[0008] The pulse wave feature extraction module is used to identify the feature points of the pulse wave, extract features, screen out the pulse wave feature vectors related to cardiac output calculation, and perform normalization processing on the pulse wave correction signal to obtain normalized feature vectors;

[0009] The dynamic signal quality monitoring module is used to monitor the quality of the normalized feature vectors in different states, and use the normalized feature vectors that meet the input conditions of the cardiac output calculation model as dynamic pulse wave feature vectors; the dynamic pulse wave feature vectors compare the quantity and value range differences of the normalized feature vectors extracted from the subject in the active state and the resting state, and output the normalized feature vectors lower than the difference threshold;

[0010] The cardiac output monitoring and evaluation module is used to dynamically and continuously monitor the subject's stroke volume and cardiac output, including: a correction variable estimation sub-module, a pulse wave feature vector estimation sub-module, and a cardiac output evaluation sub-module; the correction variable estimation sub-module corrects the variable based on the personalized feature parameter input by the subject; the pulse wave feature vector estimation sub-module sequentially calls the filtering module and the dynamic signal quality monitoring module to process the dynamic pulse wave waveform signal collected by the signal acquisition module to obtain dynamic pulse wave feature vectors; the cardiac output evaluation sub-module inputs the correction variable and the dynamic pulse wave feature vectors to obtain the stroke volume and cardiac output, and dynamically displays them on the system interface.

[0011] Preferably, in the signal acquisition module, it includes a sensor array, a pressure control sub-module, a waveform construction sub-module, and an amplification filtering and transmission sub-module;

[0012] The sensor array is used to dynamically and continuously collect the optoelectronic signals, contact pressure signals, and accelerometer signals of the peripheral artery pulse waves at different parts of the subject in the exercise state; the sensor array includes at least two optoelectronic sensors for respectively collecting the pulse waves of different peripheral artery parts, a pressure sensor for collecting the contact pressure of the corresponding parts, and an accelerometer for measuring the displacement and speed information of the subject;

[0013] The pressure control sub-module is used to apply and control the contact pressure to the peripheral artery during the acquisition process, including: an airbag for applying contact pressure to the skin, and a pressure control device for controlling the contact pressure through the airbag;

[0014] The waveform construction sub-module combines the optoelectronic signals and contact pressure signals collected by the sensor array to construct a multi-channel photoplethysmogram signal;

[0015] The amplification filtering and transmission sub-module is used to amplify and filter the multi-channel photoplethysmogram signal and accelerometer signal in sequence to obtain a dynamic pulse wave waveform signal, and transmit it to the filtering module of the host computer.

[0016] Preferably, the dynamic signal quality monitoring module includes: a signal receiving sub-module and a signal quality judging sub-module;

[0017] The signal receiving sub-module is connected to the output end of the pulse wave feature extraction module, receives the normalized feature vector and marks the signal channel to which it belongs;

[0018] The signal quality judging sub-module respectively judges whether the normalized feature vectors of different signal channels meet the input conditions of the cardiac output calculation model based on the signal quality rules, and the signal quality rules include: whether there is a lack of pulse wave feature vectors related to cardiac output calculation;

[0019] The pulse wave feature vectors related to cardiac output calculation include: pulse wave feature points, amplitude changes of each pulse wave feature point, time intervals between adjacent feature points, area under the pulse wave systolic curve, and area under the pulse wave diastolic curve;

[0020] Based on the difference judgment rule, compare the values of the pulse wave feature vectors in the active state with those in the resting state and assign scores. The differences include: amplitude change range difference, time range difference, and area range difference; the difference judgment rule includes: if the differences of the feature vectors of all channels are less than the threshold, directly use the normalized feature vector in the active state as the dynamic pulse wave feature vector and output it; if the differences of the normalized feature vectors measured at least at one part are greater than the threshold, it is necessary to reduce the influence of motion artifacts or other factors on the pulse wave signal, make the subject reduce the motion speed to the resting state and check whether the sensor wearing position has changed. When the difference is less than a certain threshold, then output the dynamic pulse wave feature vector.

[0021] Preferably, it includes: the sensor array is composed of at least two photoelectric sensors and at least one pressure sensor arranged longitudinally or transversely along the peripheral artery, an optical sensor for filtering ambient light interference, and an accelerometer worn on the subject's wrist.

[0022] Preferably, the arrangement mode of the sensor array is: at least one photoelectric sensor and at least one pressure sensor correspond to the ulnar artery part of the wrist to obtain the dynamic pulse wave waveform signal of the ulnar artery, and at least one photoelectric sensor and at least one pressure sensor correspond to the radial artery part of the wrist to obtain the dynamic pulse wave waveform signal of the radial artery.

[0023] Preferably, another arrangement of the sensor array is as follows: at least two photoelectric sensors for collecting peripheral arterial pulse waves corresponding to the ulnar artery or radial artery part of the wrist and a pressure sensor for collecting contact pressure, the sensors are arranged longitudinally along the blood vessel, and the pressure sensor is located at the central position, forming a hierarchical structure; the photoelectric signals collected by each photoelectric sensor are respectively combined with the contact pressure signal to form a two-channel dynamic pulse wave waveform signal.

[0024] Preferably, the personalized characteristic parameter generates an estimated value of the aortic cross-sectional area based on a relationship model, and the relationship model is:

[0025] (1)

[0026] S: Estimated value of aortic cross-sectional area; Age: Age; BSA: Body surface area; Height: Height; Weight: Weight; a, b, c, e are correlation coefficients, and f is a correction coefficient.

[0027] Preferably, the characteristic points of the pulse wave include: the starting point of the pulse wave, the main peak point of the pulse wave, the dicrotic wave peak point, and the end point of the pulse wave period; the pulse wave characteristic vectors related to the calculation of cardiac output include: the amplitude change of each characteristic point of the pulse wave, the time interval between adjacent characteristic points, the area under the curve during the systolic period of the pulse wave, and the area under the curve during the diastolic period of the pulse wave.

[0028] Preferably, the cardiac output calculation model is obtained based on the training sample data of the deep learning model. The sample data is divided into a training set and a test set. In the training set, the model inputs include: correction variables and pulse wave characteristic vectors related to the calculation of cardiac output, and the model output is the stroke volume. The cardiac output is calculated based on the stroke volume and heart rate. The deep learning model selects an artificial neural network model for training; the test set is used to test the trained deep learning model and evaluate it based on evaluation indicators, and change the parameters of the deep learning model until the deep learning model that meets the test and indicator requirements is used as the final cardiac output calculation model; the correction variable is the aortic cross-sectional area calculated based on the relationship model between the personalized characteristic parameter of the subject and the aortic cross-sectional area; the evaluation indicators include: model accuracy, recall rate, precision rate, etc.

[0029] Preferably, the cardiac function parameters that can also be calculated based on the cardiac output and the pulse wave signal include, but are not limited to: contractility index (CTI), ventricular ejection time (VET), heart rate (HR), continuous systolic blood pressure (SBP), continuous diastolic blood pressure (DBP), and continuous mean arterial pressure (MAP), pulse pressure variation (PPV), stroke volume variation (SVV), systemic vascular resistance (SVR), and peripheral vascular resistance index (PVR).

[0030] A dynamic non-invasive cardiac output monitoring method

[0031] Step 1, Pulse signal acquisition: Based on a sensor array and photoplethysmography, dynamically and continuously acquire multi-channel photoplethysmogram signals and accelerometer signals at different peripheral artery sites of the subject in a resting or active state, and perform filtering and amplification to obtain dynamic pulse wave waveform signals. The sensor array includes at least two photoelectric sensors for collecting pulse waves at different artery sites, and an accelerometer for measuring the displacement and velocity information of the subject.

[0032] Step 2, Pulse signal filtering: Correct the motion artifacts and baseline drift of the dynamic pulse wave waveform signal in Step 1 to generate a pulse wave correction signal.

[0033] Step 3, Pulse wave feature extraction: Identify the pulse wave feature points of the pulse wave correction signal in Step 2, extract features based on the pulse wave feature points, screen the pulse wave feature vectors related to cardiac output calculation, and perform normalization processing to obtain normalized feature vectors.

[0034] Step 4, Dynamic signal quality monitoring: Compare the quantity and value range differences of the normalized feature vectors extracted from the subject in the active state and the resting state, and based on the set difference threshold, determine whether the input conditions of the cardiac output calculation model are met to obtain dynamic pulse wave feature vectors that meet the difference threshold.

[0035] Step 5, generating correction variables: The subject inputs personalized characteristic parameters, and an estimated value of the aortic cross-sectional area is generated based on the relationship model; the relationship model is the estimated value of the aortic cross-sectional area calculated based on the age, body surface area, height, and weight of the subject:

[0036] Step 6, training the cardiac output calculation model: The sample data is divided into a training set and a test set. In the training set, based on the personalized characteristic parameters, an estimated value of the aortic cross-sectional area is calculated as the correction variable; further, the pulse wave feature vectors related to the cardiac output calculation are extracted and screened, and dynamic signal quality monitoring is performed based on Step 4. The compliant dynamic pulse wave feature vectors and the correction variable are used as inputs, the stroke volume is used as the output, and an artificial neural network is used as the pre-trained model for training; and the trained model is tested and evaluated based on the test set; the deep learning model parameters are changed until the deep learning model that meets the test and index requirements is used as the final cardiac output calculation model;

[0037] Step 7, cardiac output monitoring and evaluation: The correction variable and the dynamic pulse wave feature vectors of the subject are generated by calling Steps 1-5 and used as inputs to the cardiac output calculation model. The cardiac output calculation model of Step 6 is called to output the stroke volume and display it on the system interface.

[0038] Preferably, the step of performing motion artifact correction based on the accelerometer and the adaptive filtering algorithm is:

[0039] A1: Collect the dynamic pulse wave waveform signal of the peripheral artery and the accelerometer signal. The accelerometer signal obtains the time-displacement signal of the system in space;

[0040] A2: Signal preprocessing: Remove high-frequency noise through low-pass filtering and retain low-frequency signals; remove low-frequency drift noise through high-pass filtering and perform normalization processing;

[0041] A3: Feature extraction: Based on spectrum analysis, extract the motion characteristics of the time-displacement signals of the accelerometer in the x, y, and z axis directions; further identify the motion patterns and amplitude changes, and determine the frequency and amplitude characteristics related to motion artifacts; based on spectrum analysis, analyze the frequency and amplitude of the pulse wave signal to determine the frequency related to motion artifacts;

[0042] A4: Adaptive filtering: Use the time-displacement signal of the accelerometer as the reference signal, and automatically adjust the filter coefficients by minimizing the error between the output of the filter and the target signal to remove the motion artifacts captured by the accelerometer;

[0043] A5: Post - processing and verification: Reconstruct the artifact - removed signal to obtain the corrected pulse wave signal, ensuring that the difference in the pulse wave feature vectors of the subject collected in the motion state and the rest state is less than a certain threshold, and use evaluation metrics to quantitatively evaluate the artifact - removal effect.

[0044] Beneficial effects:

[0045] The present invention relates to a dynamic non - invasive cardiac output monitoring system and method. The system combines the photoplethysmography technology and deep learning algorithms to obtain high - quality pulse wave signals. First, the signal acquisition module continuously collects the photoplethysmogram signals of the peripheral arterial pulse waves at different sites in the static and active states and the accelerometer signals for correcting motion artifacts to obtain dynamic pulse wave signals; after screening and processing them, the features related to cardiac output calculation are extracted; since the features related to cardiac output have corrected the influence of motion artifacts, a monitoring model for cardiac output is established based on these features, thus realizing the dynamic, continuous, and long - term monitoring of cardiac output. Deep learning, especially neural networks, can effectively capture the underlying deep structure in data through multi - level data representation and non - linear transformation, providing support for effectively establishing a dynamic cardiac output monitoring model. After dynamically extracting the pulse wave features related to cardiac output calculation in the present invention, correction variables are introduced, and a module for further establishing a monitoring model for estimating cardiac output based on deep learning is included, including a dynamic signal quality monitoring module, deep learning algorithms, and a cardiac output detection and evaluation module. The dynamic signal quality monitoring module obtains the high - quality dynamic pulse wave feature vectors required for the input of the deep learning model. The deep learning algorithm takes the dynamic pulse wave feature vectors and the aortic cross - sectional area as input to the model, performs pre - training to obtain a pre - trained model for cardiac output, and obtains the pulse wave feature vectors and the aortic cross - sectional area; for the appropriate structural relationship of the stroke volume, the cardiac output is then calculated based on the stroke volume, and continuous and accurate values of the cardiac output are obtained according to the measured values of the individual.

[0046] The present invention can measure the pulse wave signal in the motion state, extract the features related to cardiac output calculation based on the pulse wave signal in the motion state, and establish a monitoring model for estimating cardiac output without the need for sphygmomanometer correction based on these features and correction features, thus realizing the dynamic, long - cycle, and accurate monitoring of cardiac output. That is, it can directly measure and collect the pulse wave of the peripheral artery without correction and without the need to maintain the measurement in the rest state as much as possible. The obtained model can more accurately identify the waveform features of the pulse wave and consider individual differences, realizing the dynamic high - precision measurement of the individual's cardiac output, which is of great significance for the long - term home rehabilitation monitoring of patients with heart failure and structural heart diseases.

[0047] In addition, the system also integrates the personalized information of the patient and the pulse wave characteristics, and constructs a deep learning model, thus truly realizing dynamic, long-term, and high-precision cardiac output monitoring.

[0048] First of all, the present invention constructs a pulse wave sensor array to solve the problem of difficult acquisition of high-precision pulse wave signals under motion states. The array includes at least two optoelectronic sensors corresponding to the relevant arterial sites on the skin, and integrates an acceleration sensor. Compared with a single sensor, the application based on the array can obtain high-fidelity pulse wave signals when fitting two or more arterial sites simultaneously. The acceleration sensor effectively reduces the position interference of the pulse wave signals caused by motion and can assist in monitoring the motion state of the subject. Based on the difference in the quality of pulse wave signals under different motion states, the present invention can perform long-term and accurate monitoring. In particular, for the filtering effect after motion artifact interference and the problem of inaccurate signal measurement caused by sensor displacement, it can significantly reduce the influence of motion and long-term use on the difference of pulse wave signals, thereby reducing the interference on the reliability of the cardiac output calculation result.

[0049] Secondly, the present invention combines photoplethysmography with a deep learning model to construct a model that can calculate cardiac output under motion states. The model is based on a deep learning model, uses the aortic cross-sectional area, which is closely related to the size of cardiac output, as an input variable. At the same time, this variable is also used as a correction variable, and combines the pulse wave characteristics and the aortic cross-sectional area as the model input, rather than directly using personalized characteristic parameters. This reduces to a certain extent the input variables that have little correlation with cardiac output and enhances the correlation between the output and the input, thereby reducing the complexity of the model.

[0050] In summary, the present invention focuses on the cardiac output monitoring technology under motion states, combines multiple technologies such as photoplethysmography, sensor array, aortic cross-sectional area estimation model, and cardiac output calculation model, realizes the accurate measurement of cardiac output in daily states, and evaluates the signal quality. This makes the calculation results reliable in motion states and long-term monitoring. The present invention improves the deficiency of the need for continuous calibration in the prior art, realizes continuous dynamic monitoring of cardiac output parameters in the true sense, and is applicable to the long-term monitoring application of home rehabilitation for patients with heart failure and structural heart diseases. Brief Description of the Drawings

[0051] Figure 1 It is a structural diagram of a dynamic non-invasive cardiac output monitoring system in the embodiment.

[0052] Figure 2 It is a structural diagram of the signal acquisition module in the embodiment.

[0053] Figure 3It is the structural diagram of the dynamic signal quality monitoring module in the embodiment.

[0054] Figure 4 It is the schematic diagram of the layout effect of the sensor array placed in the radial artery and ulnar artery in the embodiment.

[0055] Figure 5 It is the schematic diagram of the layout effect of the hierarchical sensor array in the embodiment.

[0056] Figure 6 It is the flowchart of a dynamic non-invasive cardiac output monitoring method in the embodiment. Specific implementation manners

[0057] The present invention will be further described below in conjunction with the embodiments and the accompanying drawings.

[0058] Such as Figure 1 , a dynamic non-invasive cardiac output monitoring system, comprising: a signal acquisition module, a filtering module, a pulse wave feature extraction module, a dynamic signal quality monitoring module, and a cardiac output monitoring and evaluation module;

[0059] The signal acquisition module is used to continuously acquire multi-channel photoplethysmogram signals and accelerometer signals at different peripheral artery sites in a resting or active state, and filter and amplify them to obtain dynamic pulse wave waveform signals;

[0060] The filtering module is used to correct the baseline drift and motion artifacts in the dynamic pulse wave waveform signals to obtain pulse wave corrected signals;

[0061] The pulse wave feature extraction module is used to identify pulse wave feature points, extract features, and screen pulse wave feature vectors related to cardiac output calculation from the pulse wave corrected signals, and perform normalization processing to obtain normalized feature vectors;

[0062] The dynamic signal quality monitoring module is used to monitor the quality of the normalized feature vectors in different states, and use the normalized feature vectors that meet the input conditions of the cardiac output calculation model as dynamic pulse wave feature vectors; the dynamic pulse wave feature vectors are obtained by comparing the quantity and value range differences of the normalized feature vectors extracted from the subject in the active state and the resting state, and outputting the normalized feature vectors lower than the difference threshold;

[0063] The stroke volume and cardiac output monitoring and evaluation module is used to dynamically and continuously evaluate the stroke volume and cardiac output of the subject, including: a correction variable estimation sub-module, a pulse wave feature vector estimation sub-module, and a cardiac output evaluation sub-module; the correction variable estimation sub-module uses the personalized characteristic parameters input by the subject and the estimated value of the aortic cross-sectional area as the correction variable; the pulse wave feature vector estimation sub-module sequentially calls the filtering module and the dynamic signal quality monitoring module to process the dynamic pulse wave waveform signal collected by the signal acquisition module to obtain the dynamic pulse wave feature vector; the cardiac output evaluation sub-module inputs the correction variable and the dynamic pulse wave feature vector into the cardiac output calculation model to obtain the stroke volume and cardiac output, and dynamically displays them on the system interface.

[0064] Preferably, as Figure 2 shown, in the signal acquisition module, it includes a sensor array, a pressure control sub-module, a waveform construction sub-module, and an amplification filtering and transmission sub-module;

[0065] The sensor array is used to dynamically and continuously collect the optoelectronic signals, contact pressure signals, and accelerometer signals of the peripheral artery pulse waves at different parts of the subject in the motion state; the sensor array includes at least two optoelectronic sensors that can respectively collect the pulse waves of different peripheral artery parts, a pressure sensor for collecting the contact pressure of the corresponding part, and an accelerometer for measuring the displacement and velocity information of the subject;

[0066] The pressure control sub-module is used to apply pressure to the peripheral artery during the acquisition process, including: an airbag for applying contact pressure to the skin, and a pressure control device for controlling the contact pressure through the airbag;

[0067] The waveform construction sub-module can combine the optoelectronic signals and contact pressure signals collected by the sensor array to construct a multi-channel photoplethysmogram signal;

[0068] The amplification filtering and transmission sub-module is used to amplify and filter the multi-channel photoplethysmogram signal and accelerometer signal in sequence to obtain the dynamic pulse wave waveform signal, and transmit it to the filtering module of the upper computer.

[0069] Preferably, the motion artifact is used as the motion reference signal for adaptive filtering to correct the multi-channel dynamic pulse wave waveform signals obtained from different parts.

[0070] Preferably, as Figure 3 shown, the dynamic signal quality monitoring module includes: a signal receiving sub-module and a signal quality judgment sub-module;

[0071] The signal receiving sub-module is connected to the output end of the pulse wave feature extraction module, receives the normalized feature vector, and marks the signal channel to which it belongs;

[0072] The signal quality judgment sub-module judges whether the normalized eigenvectors of different signal channels meet the input conditions of the cardiac output calculation model based on the signal quality rules. The signal quality rules include:

[0073] Whether there is a lack of pulse wave eigenvectors related to cardiac output calculation. The pulse wave eigenvectors related to cardiac output calculation include: pulse wave feature points, amplitude changes of each pulse wave feature point, time intervals between adjacent feature points, area under the curve during the systolic period of the pulse wave, and area under the curve during the diastolic period of the pulse wave;

[0074] Based on the difference judgment rule, compare the values of the pulse wave eigenvectors in the active state with those in the resting state and assign scores. The differences include: amplitude change range difference, time range difference, and area range difference. The difference judgment rule includes: if the eigenvector differences of all channels are less than the threshold, directly use the normalized eigenvector in the active state as the dynamic pulse wave eigenvector and output it; if the normalized eigenvector differences measured at least at one part are greater than the threshold, it is necessary to reduce the influence of motion artifacts or other factors on the pulse wave signal, make the subject reduce the motion speed to the resting state and check whether the sensor wearing position has changed. When the difference is less than a certain threshold, then output the dynamic pulse wave eigenvector.

[0075] Preferably, it includes: the sensor array is composed of at least two photoelectric sensors and at least one pressure sensor arranged longitudinally or transversely along the peripheral artery, an optical sensor for filtering ambient light interference, and an accelerometer worn on the subject's wrist.

[0076] Preferably, as Figure 4 shown, in one embodiment, the arrangement of the sensor array is: at least one photoelectric sensor and at least one pressure sensor correspond to the ulnar artery part of the wrist to obtain the dynamic pulse wave waveform signal of the ulnar artery, and at least one photoelectric sensor and at least one pressure sensor correspond to the radial artery part of the wrist to obtain the dynamic pulse wave waveform signal of the radial artery.

[0077] Preferably, as Figure 5 shown, in another embodiment, another arrangement of the sensor array is: at least two photoelectric sensors for collecting peripheral artery pulse waves corresponding to the ulnar artery or radial artery part of the wrist and one pressure sensor for collecting contact pressure. The sensors are arranged longitudinally along the blood vessel, and the pressure sensor is located at the central position, forming a hierarchical structure; the photoelectric signals collected by each photoelectric sensor are combined with the pressure signals respectively to form a two-channel dynamic pulse wave waveform signal.

[0078] Preferably, the personalized characteristic parameter generates an estimated value of the aortic cross-sectional area based on a relationship model, and the relationship model is as follows:

[0079] (1)

[0080] S: Estimated value of aortic cross-sectional area; Age: Age; BSA: Body surface area; Height: Height; Weight: Weight; a, b, c, e are correlation coefficients, and f is a disease correction coefficient.

[0081] In an embodiment, a correlation relationship for an aortic aneurysm patient obtained based on the above method is as follows:

[0082] (2)

[0083] Aortic aneurysm is a common structural heart disease, usually manifested as local or total segmental dilation of the aorta, resulting in an increase in the cross-sectional area.

[0084] Preferably, the characteristic points of the pulse wave include: the starting point of the pulse wave, the main peak point of the pulse wave, the dicrotic wave peak point, and the end point of the pulse wave cycle; the pulse wave feature vectors related to the calculation of cardiac output include: the amplitude changes of each characteristic point of the pulse wave, the time intervals between adjacent characteristic points, the area under the curve during the systolic period of the pulse wave, and the area under the curve during the diastolic period of the pulse wave.

[0085] Preferably, the cardiac output calculation model is obtained based on deep learning model training sample data. The sample data is divided into a training set and a test set. In the training set, the model inputs include: correction variables and pulse wave feature vectors related to the calculation of cardiac output, and the model output is the stroke volume. The cardiac output is calculated based on the stroke volume and heart rate; the deep learning model selects an artificial neural network model for training. The test set is used to test and evaluate the trained deep learning model, and the parameters of the deep learning model are changed until a deep learning model that meets the test and index requirements is used as the final cardiac output calculation model; the correction variables are based on the personalized characteristic parameters of the subject and the aortic cross-sectional area; the evaluation indexes include: model accuracy, recall rate, and precision rate.

[0086] The artificial neural network model includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer; a hidden layer and an output layer; in the artificial neural network model, the activation function can be any one of the following: Sigmoid function, Tanh function, ReLU function, Leaky ReLU function, ELU function, or PReLU function.

[0087] Preferably, in the sample data, the cardiac output and the stroke volume are derived from the Picco monitoring data, the pulse wave is derived from the pulse wave data monitored by the PPG sensor, and the personalized characteristic parameters are derived from the electronic medical record.

[0088] Preferably, the cardiac function parameters that can also be calculated based on the cardiac output and the pulse wave signal include but are not limited to: CTI, VET, HR, continuous SBP, continuous DBP, and continuous MAP, PPV, SVV, SVR, and PVR.

[0089] CTI is used to evaluate the pumping ability of the heart, and the calculation formula is:

[0090] (3)

[0091] where CO is the cardiac output, in liters per minute (L / min), and BSA is the body surface area (unit: m²).

[0092] VET is calculated based on the signal characteristics of the pulse wave signal.

[0093] HR is obtained based on the number of pulse waves collected per minute.

[0094] SBP is the maximum peak value of the pulse wave signal of the aorta, DBP is the lowest value of the aortic pulse wave signal, and MAP is an index reflecting the overall blood pressure level of the heart and vascular system, calculated based on SBP and DBP.

[0095] PPV is used to evaluate the dynamic response of the heart and blood vessels, especially the response to the respiratory cycle, and is often used to evaluate the fluid status and vascular compliance. The formula is:

[0096] (4)

[0097] where 、 and refer to the maximum value, minimum value, and average value of the systolic blood pressure, respectively.

[0098] SVV is often used to evaluate the blood volume status, especially in cases of fluid loading or dehydration. SVV can be calculated by analyzing the changes in the pulse wave signal.

[0099] (5)

[0100] where 、 、 are the maximum value, minimum value, and average value of the stroke volume, respectively.

[0101] SVR reflects the total vascular resistance and directly affects the cardiac pumping efficiency. It is usually calculated from the relationship between blood pressure and cardiac output.

[0102] Calculation formula:

[0103] (6)

[0104] Among them, MAP is the mean arterial pressure, and CO is the cardiac output, with the unit being 。

[0105] PVR is the ratio of peripheral vascular resistance to the individual's body surface area, and the calculation formula is:

[0106] (7)

[0107] Among them, SVR is the peripheral vascular resistance, and BSA is the body surface area.

[0108] Such as Figure 6 shown, a dynamic non-invasive cardiac output monitoring method

[0109] Step 1, Pulse signal acquisition: Based on a sensor array and photoplethysmography, dynamically and continuously acquire multi-channel photoplethysmogram signals and accelerometer signals from different peripheral artery sites of the subject in a resting or active state, and perform filtering and amplification to obtain dynamic pulse wave waveform signals; the sensor array includes at least two optoelectronic sensors for separately collecting pulse waves from different artery sites, and an accelerometer for measuring the displacement and velocity information of the subject;

[0110] Step 2, Pulse signal filtering: Perform motion artifact and baseline drift correction on the dynamic pulse wave waveform signal obtained in Step 1 to generate a pulse wave correction signal;

[0111] Step 3, Pulse wave feature extraction: Identify the pulse wave feature points of the pulse wave correction signal obtained in Step 2, extract features from the pulse wave correction signal based on the pulse wave feature points, and screen the pulse wave feature vectors related to cardiac output calculation, and perform normalization processing to obtain a normalized feature vector;

[0112] Step 4, Dynamic signal quality monitoring: Compare the quantity and value range differences of the normalized feature vectors extracted from the subject in the active state and the resting state, and based on a set difference threshold, determine whether it meets the input conditions of the cardiac output calculation model to obtain a dynamic pulse wave feature vector that meets the difference threshold;

[0113] Step 5, generating correction variables: The subject inputs personalized feature parameters, and an estimated value of the aortic cross-sectional area is generated based on the relationship model; the relationship model calculates the estimated value of the aortic cross-sectional area based on the subject's age, body surface area, height, and weight:

[0114] Step 6, training the cardiac output calculation model: Divide the sample data into a training set and a test set. In the training set, based on the personalized feature parameters, calculate the estimated value of the aortic cross-sectional area as the correction variable; further extract and screen the pulse wave feature vectors related to cardiac output calculation, and perform dynamic signal quality monitoring based on Step 4; use the qualified dynamic pulse wave feature vectors and correction variables as inputs, the stroke volume as the output, and an artificial neural network as the pre-trained model for training; and test and evaluate the trained model based on the test set; change the deep learning model parameters until a deep learning model that meets the test and index requirements is obtained as the final cardiac output calculation model;

[0115] Step 7, cardiac output monitoring and evaluation: Call the correction variables and dynamic pulse wave feature vectors of the subject generated in Steps 1-5 as the inputs of the cardiac output calculation model, call the cardiac output calculation model in Step 6, output the stroke volume, calculate the cardiac output based on the stroke volume and heart rate, and display it on the system interface.

[0116] Preferably, the step of performing motion artifact correction based on the accelerometer and the adaptive filtering algorithm is:

[0117] A1: Collect the dynamic pulse wave waveform signal of the peripheral artery and the accelerometer signal, and the accelerometer signal obtains the time-displacement signal of the system in space;

[0118] A2: Signal preprocessing: Perform low-pass filtering to remove high-frequency noise and retain low-frequency signals; perform high-pass filtering to remove low-frequency drift noise and perform normalization processing;

[0119] A3: Feature extraction: Extract the motion features of the time-displacement signals of the accelerometer in the x, y, and z axis directions based on spectral analysis; further identify the motion pattern and amplitude change, and determine the frequency and amplitude features related to motion artifacts; analyze the frequency and amplitude of the pulse wave signal based on spectral analysis to determine the frequency related to motion artifacts;

[0120] A4: Adaptive filtering: Use the time-displacement signal of the accelerometer as the reference signal, and automatically adjust the filter coefficients by minimizing the error between the output of the filter and the target signal to remove the motion artifacts captured by the accelerometer;

[0121] A5: Post - processing and verification: Reconstruct the artifact - removed signal to obtain the corrected pulse wave signal, ensuring that the difference in the pulse wave feature vectors of the subject collected in the motion state and in the resting state is less than a certain threshold, and quantitatively evaluate the artifact - removal effect using evaluation metrics.

[0122] Those skilled in the art can realize that the units and method steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the embodiments of the present invention.

Claims

1. A dynamic non-invasive cardiac output monitoring system, characterized in that, Including: A signal acquisition module, a filtering module, a pulse wave feature extraction module, a dynamic signal quality monitoring module, and a cardiac output monitoring and evaluation module; The signal acquisition module is used to continuously acquire multi-channel photoplethysmogram signals and accelerometer signals at different peripheral artery sites in a resting or active state, filter and amplify them to obtain dynamic pulse wave waveform signals; The filtering module is used to correct baseline drift and motion artifacts in the dynamic pulse wave waveform signals by combining accelerometer signals and an adaptive filtering algorithm to obtain pulse wave corrected signals; The pulse wave feature extraction module is used to identify pulse wave feature points, extract features, and screen and normalize pulse wave feature vectors related to cardiac output calculation from the pulse wave corrected signals to obtain normalized feature vectors; The dynamic signal quality monitoring module is used to monitor the quality of normalized feature vectors in different states, and use the normalized feature vectors that meet the input conditions of the cardiac output calculation model as dynamic pulse wave feature vectors; the dynamic pulse wave feature vectors are obtained by comparing the quantity and value range differences of the normalized feature vectors extracted from the subject in the active state and the resting state, and outputting the normalized feature vectors below the difference threshold; The cardiac output monitoring and evaluation module is used to dynamically and continuously estimate the subject's stroke volume and cardiac output, including: a correction variable estimation sub-module, a pulse wave feature vector estimation sub-module, and a cardiac output evaluation sub-module; The correction variable estimation sub-module calculates an estimated value of the aortic cross-sectional area as a correction variable based on the personalized feature parameters input by the subject; the pulse wave feature vector estimation sub-module sequentially calls the filtering module and the dynamic signal quality monitoring module to process the dynamic pulse wave waveform signals collected by the signal acquisition module to obtain dynamic pulse wave feature vectors; the cardiac output evaluation sub-module inputs the correction variable and the dynamic pulse wave feature vectors into a cardiac output calculation model trained based on a deep learning model to obtain the stroke volume and cardiac output, and dynamically displays them on the system interface.

2. The dynamic non-invasive cardiac output monitoring system according to claim 1, wherein, In the signal acquisition module, there are a sensor array, a pressure control sub-module, a waveform construction sub-module, and an amplification, filtering, and transmission sub-module; The sensor array is used to dynamically and continuously acquire photoelectric signals, contact pressure signals, and accelerometer signals of the peripheral artery pulse waves at different parts of the subject in a motion state based on photoplethysmography; the sensor array includes at least two photoelectric sensors for respectively acquiring pulse waves at different peripheral artery sites, a pressure sensor for acquiring the contact pressure at the corresponding part, and an accelerometer for measuring the displacement and velocity information of the subject, that is, body activity data; The pressure control sub-module is used to apply and control the contact pressure to the peripheral artery during the acquisition process, including: an airbag for applying contact pressure to the skin, and a pressure control device for controlling the contact pressure through the airbag; The waveform construction sub-module combines and constructs multi-channel photoplethysmogram signals based on the photoelectric signals and contact pressure signals collected by the sensor array; The amplification filtering and transmission sub-module is used to amplify and filter the photoplethysmogram signals and accelerometer signals of multiple channels in sequence to obtain dynamic pulse wave waveform signals and transmit them to the filtering module of the host computer.

3. A dynamic non-invasive cardiac output monitoring system according to claim 1, characterized in that, The motion artifact corrects the dynamic pulse wave waveform signals of multiple channels obtained from different parts based on the adaptive filtering motion reference signal of the accelerometer signal.

4. A dynamic non-invasive cardiac output monitoring system according to claim 1, characterized in that, The dynamic signal quality monitoring module includes: a signal receiving sub-module and a signal quality judgment sub-module; The signal receiving sub-module is connected to the output end of the pulse wave feature extraction module, receives the normalized feature vector and marks the signal channel to which it belongs. The signal quality judgment sub-module respectively judges whether the normalized feature vectors of different signal channels meet the input conditions of the cardiac output calculation model based on the signal quality rules. The signal quality rules include: Whether there is a lack of pulse wave feature vectors related to cardiac output calculation. The pulse wave feature vectors related to cardiac output calculation include: pulse wave feature points, the amplitude changes of each pulse wave feature point, the time intervals between adjacent feature points, the area under the curve during the systolic period of the pulse wave, and the area under the curve of the diastolic period of the pulse wave. Based on the difference judgment rule, compare the values of the pulse wave feature vectors in the active state with those in the resting state and score them. The differences include: amplitude change range difference, time range difference, and area range difference. The difference judgment rule includes: if the differences of the feature vectors of all channels are less than the threshold, directly use the normalized feature vectors in the active state as the dynamic pulse wave feature vectors and output them; if the differences of the normalized feature vectors measured at least at one part are greater than the threshold, interact with the signal acquisition module to make the subject reduce the movement speed to the resting state and check whether the sensor wearing position has changed. When the difference is less than a certain threshold, then output the dynamic pulse wave feature vectors.

5. A dynamic non-invasive cardiac output monitoring system according to claim 2, characterized in that, Including: The sensor array is composed of at least two photoelectric sensors, at least one pressure sensor arranged longitudinally or horizontally along the peripheral artery blood vessel, an optical sensor for filtering environmental light interference, and an accelerometer worn on the subject's wrist.

6. A dynamic non-invasive cardiac output monitoring system according to claim 5, characterized in that, The arrangement mode of the sensor array is: at least one photoelectric sensor and at least one pressure sensor correspond to the ulnar artery part on the skin to obtain the dynamic pulse wave waveform signal of the ulnar artery, and at least one photoelectric sensor and at least one pressure sensor correspond to the radial artery part on the skin to obtain the dynamic pulse wave waveform signal of the radial artery.

7. A dynamic non-invasive cardiac output monitoring system according to claim 2, characterized in that, The arrangement mode of the sensor array is: at least two photoelectric sensors for collecting peripheral artery pulse waves corresponding to the ulnar artery or radial artery part on the skin and one pressure sensor for collecting contact pressure. The sensors are arranged longitudinally along the blood vessel, and the pressure sensor is located at the central position, forming a sandwich structure; the photoelectric signals collected by each photoelectric sensor are combined with the contact pressure signals respectively to form two-channel dynamic pulse wave waveform signals.

8. A dynamic non-invasive cardiac output monitoring system according to claim 2, wherein, The personalized feature parameter calculates the estimated value of the aortic cross-sectional area based on the relationship model. The relationship model is constructed by fitting the linear relationship between age, body surface area, height, weight, and the estimated value of the aortic cross-sectional area: (1) S: Estimated value of aortic cross-sectional area; Age: Age; BSA: Body surface area; Height: Height; Weight: Weight; a, b, c, e are correlation coefficients, and f is a correction coefficient.

9. A dynamic non-invasive cardiac output monitoring system according to claim 2, characterized in that, The characteristic points of the pulse wave include: the starting point of the pulse wave, the main peak point of the pulse wave, the dicrotic wave peak point, and the end point of the pulse wave period; the pulse wave characteristic vectors related to the calculation of cardiac output include: the amplitude changes of each characteristic point of the pulse wave, the time intervals between adjacent characteristic points, the area under the curve during the systolic period of the pulse wave, and the area under the curve during the diastolic period of the pulse wave.

10. A dynamic non-invasive cardiac output monitoring system according to claim 1 or 2, characterized in that, The cardiac output calculation model is obtained based on the training sample data of the deep learning model. The sample data is divided into a training set and a test set. In the training set, the model inputs include: correction variables and pulse wave characteristic vectors related to the calculation of cardiac output. The model output is the stroke volume. The cardiac output is calculated based on the stroke volume and heart rate. The deep learning model selects an artificial neural network model for training; the test set is used to test the trained deep learning model and evaluate it based on evaluation metrics, and the parameters of the deep learning model are changed until the deep learning model that meets the test and metric requirements is used as the final cardiac output calculation model; the correction variable is the aortic cross-sectional area calculated based on the relationship model between the personalized characteristic parameters of the subject and the aortic cross-sectional area; the evaluation metrics include: model accuracy, recall rate, precision rate, and F1 value.

11. A dynamic non-invasive cardiac output monitoring system according to claim 10, characterized in that, In the sample data, the cardiac output and stroke volume are from Picco monitoring data, the pulse wave is from the pulse wave data monitored by the PPG sensor, and the personalized characteristic parameters are from the electronic medical record.

12. A dynamic non-invasive cardiac output monitoring system according to claim 11, characterized in that, The cardiac function parameters that can also be calculated based on the cardiac output and pulse wave signal include but are not limited to: cardiac contractility index, ventricular ejection time, heart rate, continuous systolic blood pressure, continuous diastolic blood pressure and continuous mean blood pressure, pulse pressure variability, stroke volume variability, vascular peripheral resistance, and vascular peripheral resistance index.

13. A dynamic non-invasive cardiac output monitoring method based on the system according to any one of claims 1-12, characterized in that Step 1, Pulse signal acquisition: Dynamically and continuously collect multi-channel photoplethysmogram signals and accelerometer signals of different peripheral artery sites of the subject at rest or in an active state based on the sensor array and photoplethysmography, and perform filtering and amplification to obtain dynamic pulse wave waveform signals. The sensor array includes at least two photoelectric sensors for separately collecting pulse waves of different artery sites, and an accelerometer for measuring the displacement and velocity information of the subject; Step 2, Pulse signal filtering: Combine the accelerometer signal and the adaptive filtering algorithm to correct the motion artifacts and baseline drift of the dynamic pulse wave waveform signal in Step 1 to generate a pulse wave correction signal; Step 3, Pulse wave feature extraction: Perform pulse wave characteristic point recognition, feature extraction on the pulse wave correction signal in Step 2, screen the pulse wave characteristic vectors related to the calculation of cardiac output, and perform normalization processing to obtain a normalized feature vector; Step 4, Dynamic signal quality monitoring: Compare the differences in the quantity and value range of the normalized feature vectors extracted from the subject in the active state and the resting state, and judge whether the input conditions of the cardiac output calculation model are met based on the set difference threshold, and obtain the dynamic pulse wave feature vectors that meet the difference threshold; Step 5, Generate correction variables: The subject inputs personalized feature parameters to generate an estimated value of the aortic cross-sectional area based on the relationship model; the relationship model is a calculation model for linearly fitting the estimated value of the aortic cross-sectional area based on the subject's age, body surface area, height, and weight; Step 6, Train the cardiac output calculation model: Divide the sample data into a training set and a test set. In the training set, calculate the estimated value of the aortic cross-sectional area as a correction variable based on the personalized feature parameters; further extract and screen the pulse wave feature vectors related to cardiac output calculation and perform dynamic signal quality monitoring based on Step 4. Use the qualified dynamic pulse wave feature vectors and correction variables as inputs, the stroke volume as the output, and the artificial neural network as the pre-training model for training; and test and evaluate the trained model based on the test set; Change the deep learning model parameters until the deep learning model that meets the test and index requirements is used as the final cardiac output calculation model; Step 7, Cardiac output monitoring and evaluation: Call Steps 1-5 to generate the subject's correction variables and dynamic pulse wave feature vectors as inputs to the cardiac output calculation model, call the cardiac output calculation model in Step 6, output the stroke volume, calculate the cardiac output based on the stroke volume and heart rate, and display it on the system interface.

14. A dynamic non-invasive cardiac output monitoring method according to claim 13, characterized in that The steps for motion artifact correction based on the accelerometer and the adaptive filtering algorithm are as follows: A1: Collect the dynamic pulse wave waveform signal of the peripheral artery and the accelerometer signal. The accelerometer signal obtains the time-displacement signal of the system in space; A2: Signal preprocessing: Low-pass filtering: Remove high-frequency noise and retain low-frequency signals; High-pass filtering: Remove low-frequency drift noise and perform normalization processing; A3: Feature extraction: Based on spectral analysis, extract the motion features of the time-displacement signals of the accelerometer in the x, y, and z axis directions; and further identify the motion pattern and amplitude change, and determine the frequency and amplitude features related to motion artifacts; Analyze the frequency and amplitude of the dynamic pulse wave waveform signal based on spectral analysis to determine the frequency related to motion artifacts; A4: Adaptive filtering: Use the time-displacement signal of the accelerometer as the reference signal, and automatically adjust the filter coefficients by minimizing the error between the output of the filter and the target signal to remove the motion artifacts captured by the accelerometer; A5: Post-processing and verification: Reconstruct the artifact-removed signal to obtain the pulse wave correction signal to ensure that the difference in the pulse wave feature vectors collected from the subject in the motion state and the resting state is less than a certain threshold, and use evaluation indicators to quantitatively evaluate the artifact-removing effect.

Images