Method and related device for continuous blood pressure monitoring based on multiple physiological signals and electrical network

Abstract

The application discloses a continuous blood pressure monitoring method based on multiple physiological signals and an electric network and related equipment. An electric network model is constructed in advance based on a first resistance, a second resistance, a capacitance and an inductance; a physiological signal set of a user is acquired; a model parameter combination of the electric network model is predicted according to signal characteristics of the physiological signal set through a prediction model; the prediction model parameter combination is applied to the electric network model, and a corresponding simulation blood pressure waveform is simulated and generated, and a blood pressure monitoring result of the user is determined according to the simulation blood pressure waveform. The application extracts signal characteristics related to blood pressure changes from multiple physiological signals, predicts model parameters of the electric network model, and realizes non-invasive continuous blood pressure monitoring according to the relationship between the model parameters and blood pressure changes. The multiple physiological signals can fully utilize the advantages and complementary characteristics of different physiological signals, provide more comprehensive information, and the electric network model can simulate the changes of the cardiovascular system, thereby enhancing the accuracy and reliability of continuous blood pressure monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of continuous blood pressure detection, and in particular to a continuous blood pressure monitoring method and related equipment based on multiple physiological signals and an electrical network model. Background Art

[0002] Blood pressure, a crucial physiological indicator of the human circulatory system, plays a vital role in maintaining cardiovascular balance. Accurate and timely blood pressure monitoring is crucial for preventing hypertension and related cardiovascular diseases.

[0003] Currently, common continuous non-invasive blood pressure monitoring methods encompass a variety of technologies, including those based on pressure sensors, ultrasound, and photoplethysmography. Continuous non-invasive blood pressure monitoring can more comprehensively and conveniently reflect an individual's daily blood pressure changes, blood pressure levels over time, blood pressure control status, and blood pressure trends after specific training sessions.

[0004] However, existing noninvasive continuous blood pressure monitoring methods generally use a single signal, which makes it difficult to fully reflect the status of the cardiovascular system, thus affecting the accuracy of continuous blood pressure monitoring. In addition, existing methods still have shortcomings in clinical physiological interpretation and are limited in their ability to reflect cardiovascular system and hemodynamic information, making it difficult to reveal the impact of cardiovascular system changes on blood pressure.

[0005] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a continuous blood pressure monitoring method and related equipment based on multiple physiological signals and electrical networks in response to the above-mentioned defects of the existing technology, aiming to solve the problem that the existing non-invasive continuous blood pressure detection methods generally use a single signal, which is difficult to fully reflect the condition of the cardiovascular system, thereby affecting the accuracy of continuous blood pressure monitoring.

[0007] The technical solutions adopted by the present invention to solve the problem are as follows:

[0008] In a first aspect, an embodiment of the present invention provides a continuous blood pressure monitoring method based on multiple physiological signals and an electrical network, the method comprising:

[0009] Obtaining an electrical network model pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; wherein a current input corresponding to the electrical network model is determined based on a cardiac cycle time, a cardiac systolic ejection time, and a cardiac output; and model parameters of the electrical network model include variable heart rate, stroke volume, a first resistance value, and a capacitance value, and constant cardiac systolic ejection time, a second resistance value, and an inductor value;

[0010] Acquire a physiological signal set of a target user; wherein the physiological signal set includes a plurality of physiological signals of different types;

[0011] Extracting features from the physiological signal set, and predicting model parameters of the electrical network model based on the signal features of the physiological signal set using a preset prediction model to obtain a prediction model parameter combination;

[0012] The prediction model parameter combination is applied to the electrical network model, and a corresponding target simulated blood pressure waveform is simulated to generate a corresponding blood pressure monitoring result of the target user is determined based on the target simulated blood pressure waveform.

[0013] In a second aspect, an embodiment of the present invention provides a continuous blood pressure monitoring system based on multiple physiological signals and an electrical network, the system comprising:

[0014] An electrical network model, wherein the electrical network model is pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; a current input corresponding to the electrical network model is determined based on a cardiac cycle time, a cardiac systolic ejection time, and a cardiac output; model parameters of the electrical network model include variable heart rate, stroke volume, a first resistance value, and a capacitance value; and constant cardiac systolic ejection time, a second resistance value, and an inductor value;

[0015] A signal acquisition module, configured to acquire a physiological signal set of a target user; wherein the physiological signal set includes a plurality of physiological signals of different types;

[0016] a signal processing module, configured to extract features from the physiological signal set, and predict model parameters of the electrical network model based on the signal features of the physiological signal set using a preset prediction model to obtain a prediction model parameter combination;

[0017] The data simulation module is used to apply the prediction model parameter combination to the electrical network model, and simulate and generate a corresponding target simulated blood pressure waveform, and determine the blood pressure monitoring result corresponding to the target user based on the target simulated blood pressure waveform.

[0018] In a third aspect, an embodiment of the present invention provides a terminal comprising a memory and one or more processors; the memory stores one or more programs; the programs include instructions for executing any of the continuous blood pressure monitoring methods based on multiple physiological signals and electrical networks as described above; and the processor is used to execute the programs.

[0019] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium on which multiple instructions are stored, characterized in that the instructions are suitable for being loaded and executed by a processor to implement the steps of any of the above-mentioned continuous blood pressure monitoring methods based on multiple physiological signals and electrical networks.

[0020] Beneficial effects of the present invention: Embodiments of the present invention extract signal features related to blood pressure changes from multiple physiological signals, predict model parameters of an electrical network model, and implement noninvasive and continuous blood pressure monitoring based on the relationship between model parameters and blood pressure changes. Multiple physiological signals can fully leverage the strengths and complementary properties of different physiological signals to provide more comprehensive information, while the electrical network model can simulate changes in the cardiovascular system, thereby enhancing the accuracy and reliability of continuous blood pressure monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 The figure is a flowchart of the steps of a continuous blood pressure monitoring method based on multiple physiological signals and electrical networks provided by an embodiment of the present invention.

[0023] Figure 2 It is a structural diagram of an electrical network model provided by an embodiment of the present invention.

[0024] Figure 3 It is a schematic diagram of some morphological features provided by an embodiment of the present invention.

[0025] Figure 4 Schematic diagram of a portion of a simulated blood pressure waveform provided by an embodiment of the present invention.

[0026] Figure 5 It is a schematic diagram of matching between a simulation signal and a real signal provided by an embodiment of the present invention.

[0027] Figure 6 Schematic diagram of a module of a continuous blood pressure monitoring system based on multiple physiological signals and electrical networks provided by an embodiment of the present invention.

[0028] Figure 7 This is a principle block diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The present invention discloses a continuous blood pressure monitoring method and related equipment based on multiple physiological signals and an electrical network. To make the objectives, technical solutions, and effects of the present invention more clear and explicit, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention.

[0030] It will be understood by those skilled in the art that, unless expressly stated otherwise, the singular forms "a", "an", "said" and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the description of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. The term "and / or" used herein includes all or any units and all combinations of one or more associated listed items.

[0031] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0032] In response to the above-mentioned shortcomings of the prior art, the present invention provides a continuous blood pressure monitoring method based on multiple physiological signals and an electrical network. The method obtains an electrical network model pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; wherein the current input corresponding to the electrical network model is determined based on a cardiac cycle time, cardiac systolic ejection time, and cardiac output; the model parameters of the electrical network model include variable heart rate, stroke volume, first resistance value, and capacitance value, and constant cardiac systolic ejection time, second resistance value, and inductance value; obtains a physiological signal set of a target user; wherein the physiological signal set includes several physiological signals of different types; extracts features from the physiological signal set, and predicts the model parameters of the electrical network model based on the signal features of the physiological signal set using a preset prediction model to obtain a prediction model parameter combination; applies the prediction model parameter combination to the electrical network model, and simulates and generates a corresponding target simulated blood pressure waveform, and determines the blood pressure monitoring result corresponding to the target user based on the target simulated blood pressure waveform. The present invention extracts signal features related to blood pressure changes from multiple physiological signals, predicts the model parameters of the electrical network model, and realizes non-invasive continuous blood pressure monitoring based on the relationship between the model parameters and blood pressure changes. Multiple physiological signals can make full use of the advantages and complementary characteristics of different physiological signals to provide more comprehensive information, and the electrical network model can simulate changes in the cardiovascular system, thereby enhancing the accuracy and reliability of continuous blood pressure monitoring.

[0033] like Figure 1 As shown, the method specifically includes the following steps:

[0034] Step S100, obtaining an electrical network model pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; wherein the current input corresponding to the electrical network model is determined based on a cardiac cycle time, a cardiac systolic ejection time, and a cardiac output; the model parameters of the electrical network model include a variable heart rate, a stroke volume, a first resistance value, and a capacitance value, and a constant cardiac systolic ejection time, a second resistance value, and an inductor value.

[0035] Specifically, the electrical network model is a mathematical model that reflects the physiological characteristics of the cardiovascular system. The model can mathematically link the blood flow in the arterial blood vessels and the blood pressure. The electrical network model pre-constructed in this embodiment is composed of two resistors, a capacitor and an inductor. The model parameters of the electrical network model are divided into two categories: heart rate (HR), stroke volume (SV), the first resistance value (R1) and the capacitance value (C) are variable model parameters; the cardiac systolic ejection time (T s ), the second resistance value (R2) and the inductance value (L) are preset constant model parameters. The current input of the electrical network model will be based on the cardiac cycle time (T c ), cardiac systolic ejection time (T s ), cardiac output (CO).

[0036] For example, the electrical network model has the following parameters: T s : cardiac systolic ejection time; T c : The duration of a cardiac cycle, equivalent to heart rate (HR); CO: Cardiac output, combined with HR to obtain stroke volume (SV); R1: The value of the first resistor; R2: The value of the second resistor; C: The value of the capacitor; L: The value of the inductor. Among the above parameters, the parameter that needs to be preset as a constant is T s , R2, and L can be preset to 0.25, 0.05, and 0.005, respectively.

[0037] The current input of the circuit is I(t), and its change expression is shown in the following formula, where T c is the time of one cardiac cycle, T s is the cardiac contraction ejection time, CO is the cardiac output (i.e. the amount of blood ejected per minute), and n is the nth cardiac cycle.

[0038]

[0039]

[0040] The electrical network model can be implemented in Simulink simulation, or its state equation can be derived based on the principle and implemented through programming as shown in the following formula:

[0041]

[0042] Where P(t) is the voltage.

[0043] This embodiment is simulated using Simulink.

[0044] In another implementation, electrical network model parameters that are inconsistent with those in this embodiment may be preset.

[0045] In one implementation, the first resistor and the inductor are connected in parallel to form a first parallel module; the second resistor and the capacitor are connected in parallel to form a second parallel module; and the first parallel module and the second parallel module are connected in series.

[0046] Specifically, this embodiment uses a first resistor, a second resistor, a capacitor, and an inductor to design a four-element second-order electrical network model. Figure 2 As shown, a first resistor R1 is connected in parallel with an inductor L, a second resistor R2 is connected in parallel with a capacitor C, and then the two components are connected in series. This embodiment uses a four-element electrical network for modeling, which can effectively simulate the cardiovascular system, provide a deeper understanding of the impact of changes in the cardiovascular system on blood pressure changes, and explore the impact of changes in different parameter values ​​of the electrical network model on blood pressure changes, further analyzing the differences in model parameters for users with different blood pressure conditions.

[0047] In one implementation, the analogy between the model parameters of the four-element electrical network model and the physiological characteristics of the cardiovascular system is: total peripheral resistance is analogous to resistance; arterial compliance is analogous to capacitance; blood inertia is analogous to inductance; blood flow from the ventricle to the artery is analogous to time-varying current; and arterial pressure wave is analogous to time-varying voltage.

[0048] Specifically, the elasticity of human vascular walls and the peripheral resistance of blood vessels have regular influences on the formation of physiological signals. The characteristics of these multiple physiological signals can, to a certain extent, reflect the numerical changes in the electrical network model parameters of resistance, capacitance, and inductance. Therefore, the electrical network model used in this embodiment can simulate the cardiovascular system, effectively modeling it, making the model highly interpretable and enabling accurate and reliable continuous blood pressure monitoring.

[0049] In another implementation, different electrical network models may be designed, such as five-element and six-element electrical network models.

[0050] Step S200: Acquire a physiological signal set of a target user; wherein the physiological signal set includes several physiological signals of different types.

[0051] Specifically, multiple physiological signals can fully leverage the strengths and complementary properties of different physiological signals to provide more comprehensive information. Therefore, for users requiring continuous blood pressure monitoring, this embodiment first requires obtaining several different types of physiological signals from the user, thereby generating a physiological signal set. Subsequent continuous blood pressure monitoring using this physiological signal set can effectively enhance the accuracy and reliability of continuous blood pressure monitoring.

[0052] In one implementation, after obtaining the physiological signal set of the target user, the method further includes:

[0053] Each physiological signal in the physiological signal set is preprocessed to remove noise; wherein the preprocessing includes: processing low-frequency interference in each physiological signal through wavelet transform, and processing high-frequency interference in each physiological signal through a filter of a preset frequency.

[0054] Specifically, since the acquired physiological signals are often accompanied by a certain amount of noise, preprocessing is crucial. Noise is mainly divided into low-frequency and high-frequency interference: low-frequency interference is mainly manifested as baseline drift of the waveform, which is mainly derived from the low-frequency signal generated by human breathing; high-frequency interference is mainly manifested as small-amplitude fluctuations on the waveform signal, and its sources include the myoelectric noise generated by the normal physiological activities of human tissue and the electrical signal interference of the acquisition equipment itself. In the process of signal preprocessing, it is necessary to reduce baseline drift and power frequency interference as much as possible while retaining the waveform characteristics of the original signal to improve the quality of the signal. For low-frequency noise, such as respiratory movement, which can cause changes in the volume of the chest and abdomen, thereby causing changes in the signal, this embodiment uses wavelet transform for processing to reduce the impact of respiratory interference on the signal. For high-frequency noise, this embodiment uses a preset frequency filter for processing, such as a 4th-order Butterworth filter with a cutoff frequency of 10Hz for processing to eliminate high-frequency noise in the signal.

[0055] Step S300 : extracting features from the physiological signal set, and predicting model parameters of the electrical network model according to the signal features of the physiological signal set using a preset prediction model to obtain a prediction model parameter combination.

[0056] Specifically, this embodiment pre-builds a prediction model using a machine learning algorithm. The input data for this prediction model are the signal characteristics of multiple physiological signals, and the output data are the model parameter combinations predicted based on these signal characteristics. This embodiment uses the prediction model to obtain specific parameters of the electrical network model from the characteristics of multiple physiological signals, thereby enabling accurate continuous blood pressure monitoring using the electrical network model.

[0057] In one implementation, extracting features from the physiological signal set includes:

[0058] Dividing each physiological signal in the physiological signal set according to the cardiac cycle to obtain signal segments corresponding to each physiological signal;

[0059] Each of the signal segments is divided into multiple rounds for feature extraction, wherein each of the signal segments in the same round corresponds to the same cardiac cycle.

[0060] Specifically, this embodiment divides each physiological signal into signal segments according to the cardiac cycle, and extracts signal features from the signal segments divided based on the cardiac cycle, so that the electrical network model can update the model parameters according to the cardiac cycle beat, thereby realizing non-invasive blood pressure monitoring according to the cardiac cycle.

[0061] In one implementation, the physiological signal set includes an ECG signal, an ICG signal, and a PPG signal; and the prediction model is used to:

[0062] predicting the heart rate of the electrical network model based on the signal characteristics of the ECG signal;

[0063] predicting the beat-per-beat output of the electrical network model based on signal characteristics of the ICG signal;

[0064] A first resistance value and a first capacitance value of the electrical network model are predicted according to a signal characteristic of the PPG signal.

[0065] Specifically, multiple physiological signals can fully reflect changes in the body's physiological information and carry information about blood pressure changes. The multiple physiological signals used in this embodiment, namely, the physiological signals collectively include: electrocardiogram (ECG), photoplethysmography (PPG), and intracardiac impedance (ICG). ECG signals can reflect the heart's electrical activity and the overall rhythm of the heartbeat; PPG signals can reflect the condition of peripheral arteries and blood flow; and ICG signals can reflect changes in thoracic (aortic) blood flow and blood volume changes. These three physiological signals carry different information, but their complementarity can more comprehensively reflect the condition of the human body, especially the cardiovascular system. Combining these three signals can more accurately monitor changes in blood pressure. In practical applications, ECG signals are primarily used to segment cardiac cycles. Since the duration of a cardiac cycle is equivalent to heart rate, signal features can be extracted from ECG signals to set the heart rate in the electrical network model parameters. ICG signals are primarily used to analyze cardiac function and calculate blood flow parameters. Stroke volume, a parameter of the electrical network model, can be calculated based on ICG signals. The PPG signal mainly reflects the conditions of peripheral arterial blood vessels and blood flow, so features can be extracted from the PPG signal to set the first resistance value and capacitance value in the electrical network model parameters.

[0066] For example, after completing the preprocessing of each physiological signal, feature point extraction will be performed:

[0067] For ECG signals, we mainly focus on its R peak for segmentation of cardiac cycles.

[0068] For ICG signals, we focus on points C, B, and X. These characteristic points are crucial for analyzing cardiac function and calculating blood flow parameters. SV can be obtained from the ICG signal as shown in the following formula:

[0069]

[0070] LVET=t X -t B , (6);

[0071] Where, LVET is left ventricular ejection time; h c is the amplitude at point C; h B is the amplitude at point B; t X is the time of point X; t B The moment of point B.

[0072] For PPG signals, we focus on one or more of the following: morphological, derivative, statistical, sequence, and frequency domain features. Compared to traditional methods that focus on waveform and morphological features of physiological signals, we can uncover deeper features from multiple layers, including morphological, derivative, statistical, sequence, and frequency domain features.

[0073] Morphological features may include one or more of the following features: amplitude at the starting point, amplitude at the peak point, duration of systole, duration of diastole, amplitude and duration at the time when systole duration is 10%, 25%, 33%, 50%, 66%, 75%, amplitude and duration at the time when diastole duration is 10%, 25%, 33%, 50%, 66%, 75%, etc. 38 features, some of which are as follows. Figure 3 shown.

[0074] The derivative features may include one or more of the following features: the time at which the first-order derivative has a maximum value and the amplitude in the original signal, the time at which the second-order derivative has a maximum value and the amplitude in the original signal, etc. 12 features.

[0075] Statistical features may include one or more of the following features: signal mean, variance, skewness, and other nine features.

[0076] The sequence features may include one or more of the following features: 15 features such as the approximate entropy and sample entropy 4 seconds before and after the period.

[0077] The frequency domain features may include one or more of the following features: the sum of spectrum energy, maximum spectrum energy, and other 14 features.

[0078] There are 88 features in five categories.

[0079] In another implementation, physiological signal feature extraction that is not completely consistent with the present invention may be performed.

[0080] In one implementation, the prediction model is optimized in advance using training data, and a method for generating the training data includes:

[0081] Predetermine the values ​​corresponding to the constant model parameters respectively, and set the parameter grid intervals for the variable model parameters respectively to perform traversal and combination, and obtain several model parameter combinations;

[0082] Simulating the electrical network model according to each of the model parameter combinations to obtain simulated blood pressure waveforms corresponding to each of the model parameter combinations;

[0083] Acquire a plurality of real physiological signal sets and real blood pressure waveforms corresponding to the real physiological signal sets;

[0084] Performing parameter matching on each of the real blood pressure waveforms and each of the simulated blood pressure waveforms, and determining a correspondence between each of the real physiological signal sets and each of the model parameter combinations based on the matching results;

[0085] A number of training data are generated based on each of the real physiological signal sets and each of the model parameter combinations, wherein each of the training data includes a pair of real physiological signal sets and model parameter combinations with a corresponding relationship, the real physiological signal sets are input into the prediction model as training samples, and the model parameter combinations are used as target values ​​corresponding to the training samples to evaluate the accuracy of the predicted values ​​output by the prediction model.

[0086] Specifically, the prediction model must first learn the complex mapping relationship between real physiological signal sets and model parameter combinations through a large amount of training data, so that the trained prediction model can accurately predict the corresponding model parameter combination based on the input physiological signal set. In practical application, the values ​​of each constant model parameter are first set. Then, the parameter grid intervals of each variable model parameter are set and traversed to obtain multiple model parameter combinations. The electrical network model is simulated for each model parameter combination to generate simulated blood pressure waveforms. The number of simulated blood pressure waveforms generated is consistent with the total number of model parameter combinations. Then, a large number of real physiological signal sets (ICG signals + ECG signals) and the real blood pressure waveforms corresponding to each real physiological signal set are obtained. For each real physiological signal set, the simulated blood pressure waveform that most closely matches the real physiological signal set is matched by comparing key parameters such as systolic blood pressure, diastolic blood pressure, heart rate, and stroke volume. Because each simulated blood pressure waveform has a one-to-one correspondence with each model parameter combination, the matching results can indirectly determine the corresponding relationship between each real physiological signal set and each model parameter combination, and then multiple training data are generated based on this correspondence. Each training data consists of a set of real physiological signals (input into the prediction model as training samples) and a corresponding combination of model parameters (used as target values / true labels to evaluate the prediction accuracy of the prediction model).

[0087] For example, the constant model parameter T is pre-set s The values ​​of , R2 and L can be preset to 0.25, 0.05 and 0.005 respectively. The parameter grid intervals of other model parameters are set to traverse and combine, such as HR is set between 50-120 with a step size of 10; SV is set between 50-100 with a step size of 10; R1 and C are set between 0.5-4 with a step size of 0.01. Simulink simulation is performed on the electrical network model under the above model parameter combination to generate simulated blood pressure waveforms. Some example waveforms are as follows Figure 4As shown. After generating simulated blood pressure waveforms under many different model parameter combinations, the following information contained in these simulated blood pressure waveforms is obtained: systolic pressure, diastolic pressure, heart rate and stroke volume. For real physiological signals, the stroke volume in each cardiac cycle is obtained from the ICG signal, the heart rate is obtained from the ECG signal corresponding to the cardiac cycle, and the systolic and diastolic pressure values ​​are obtained from the blood pressure waveform corresponding to the cardiac cycle. From the numerous simulated blood pressure waveforms, a simulated blood pressure waveform with the parameters closest to the real physiological signal is found to achieve matching between the real blood pressure waveform and the simulated blood pressure waveform (such as Figure 5 Using this method, real physiological signals (i.e., ECG, continuous blood pressure, and ICG signals) can be mapped one-to-one to the model parameter combinations of the electrical network model to achieve data matching. The prediction model can use the results of data matching to learn the mapping from multiple physiological signals to the model parameter combinations of the electrical network model.

[0088] In one implementation, all data are divided at the sample level, with the ratio of training set, validation set, and test set being 8:1:1.

[0089] In one implementation, the prediction model is constructed using a gradient boosted decision tree, and the model optimization method of the prediction model includes:

[0090] For each round of the multiple rounds of iterative optimization, a training sample of the round is obtained, and a prediction value is generated according to the training sample through all decision trees of the prediction model;

[0091] Calculate the model residual corresponding to the round according to the predicted value and the corresponding target value;

[0092] Determining whether the model residual converges to the optimization target; if the model residual does not converge to the optimization target, constructing a new decision tree based on the model residual, and determining the splitting points of the sub-leaves of the decision tree according to a greedy algorithm;

[0093] Obtain the next round of training samples, and continue to execute the step of generating prediction values ​​according to the training samples through all decision trees of the prediction model until the model residual converges to the optimization target.

[0094] Specifically, the prediction model of this embodiment is constructed using a gradient boosting decision tree, which is an iterative decision tree algorithm. Each round of iteration adds a new decision tree to fit the residual of the previous round, and uses the new and old trees together for prediction. The construction of the decision tree is based on a greedy algorithm, recursively selecting the split point to maximize the gain. Different sub-leaves in the decision tree correspond to different prediction values, and the split point is to split a sub-leaves to obtain a more detailed prediction value. Multiple rounds of iterative optimization are performed until the residual of the prediction model meets the standard, which means that the prediction accuracy of the prediction model meets the standard, and the training optimization of the prediction model is completed.

[0095] For example, the prediction model is based on the gradient boosting tree. The specific steps of building / optimizing the algorithm are as follows:

[0096] First, initialize the model prediction value and set the initialization value to the mean of the target value of the training dataset:

[0097]

[0098] Where, is the initialization value; n is the total amount of training data; y i is the target value of the i-th training data.

[0099] Then, the prediction model is optimized using multiple iterations. Each iteration adds a new tree to fit the residual of the previous round. In each iteration, the residual needs to be calculated, that is, the difference between the target value (such as the actual R1 or C) and the predicted value output by the current prediction model:

[0100]

[0101] Where, is the residual of round t.

[0102] A new decision tree is constructed based on the residuals, so that the tree output best fits the residuals. Decision trees are constructed using a greedy algorithm, recursively selecting split points to maximize gain. Different sub-leaves in a decision tree correspond to different prediction values. A split point is where a sub-leave is split to obtain a more detailed prediction value. For example, a prediction value of 1 for a sub-leave can be split into 0.5 and 1.5.

[0103] The gain formula of the split point is:

[0104]

[0105] Among them, G and H are the gradient and the sum of the second-order derivative of the leaf node respectively, G L and H L is the sum of the gradient and second-order derivative of the left child node, G R and H Ris the sum of the gradient and the second-order derivative of the right child node, λ is the regularization parameter, and γ is the penalty term for the leaf node.

[0106] Next, use the learning rate to update the model's predictions:

[0107]

[0108] Among them, f t is the newly constructed tree in round t; η is the learning rate, which controls the contribution of each tree to the final prediction value.

[0109] The loss function that the prediction model minimizes in each round of iteration is the objective function, which includes the training error Obj(θ) and the regularization term Ω(f):

[0110]

[0111] Where l is the loss function (e.g., squared error); Ω is the regularization term used to control model complexity; T is the number of leaf nodes in the tree; ω j is the weight of the leaf node; γ and λ are hyperparameters used to control the degree of regularization.

[0112] Finally, the parameters are continuously updated in the iteration so that the loss function is continuously reduced, thus achieving the final construction / optimization of the prediction model.

[0113] In another implementation, other machine learning algorithms or deep learning algorithms may be used to predict the model parameter combination of the electrical network model.

[0114] Step S400: Apply the prediction model parameter combination to the electrical network model, simulate and generate a corresponding target simulated blood pressure waveform, and determine the blood pressure monitoring result corresponding to the target user based on the target simulated blood pressure waveform.

[0115] Specifically, the predicted parameters of the electrical network model are applied to the electrical network model, and simulation software is used to perform simulation to obtain the corresponding simulated blood pressure waveform. The user's blood pressure monitoring results, such as systolic pressure and diastolic pressure, can be obtained through the simulated blood pressure waveform, thereby realizing non-invasive continuous blood pressure monitoring.

[0116] For example, during the model testing phase, the parameters of the electrical network model predicted by the test set can be applied to the electrical network model, and Simulink can be used for simulation to generate a simulated blood pressure waveform to obtain the systolic and diastolic pressures.

[0117] The advantages of the present invention are:

[0118] 1. The present invention proposes to fuse multiple physiological signals, which helps to give full play to the complementarity between different signals and provide more comprehensive and accurate information related to the cardiovascular system and blood pressure changes. It makes up for the limitations of the information provided by a single signal in blood pressure monitoring, thereby enhancing the accuracy and reliability of blood pressure monitoring.

[0119] 2. The present invention proposes to use a four-element electrical network model as a model for continuous blood pressure monitoring. Compared with the existing black box model methods based on deep learning or machine learning, it is more physiologically interpretable. Specifically, the electrical network model can well model the changes in the cardiovascular system and can intuitively reflect the process of blood pressure changes.

[0120] 3. The present invention extracts signal features from signal segments divided according to the cardiac cycle, so that the electrical network model can update parameters according to the cardiac cycle beat, thereby realizing non-invasive blood pressure monitoring according to the cardiac cycle.

[0121] Based on the above embodiments, the present invention also provides a continuous blood pressure monitoring system based on multiple physiological signals and electrical networks, such as Figure 6 As shown, the system includes:

[0122] An electrical network model 01 is pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; a current input corresponding to the electrical network model is determined based on a cardiac cycle time, a cardiac systolic ejection time, and a cardiac output; model parameters of the electrical network model include variable heart rate, stroke volume, a first resistance value, and a capacitance value; and constant cardiac systolic ejection time, a second resistance value, and an inductor value.

[0123] Signal acquisition module 02, used to acquire a physiological signal set of a target user; wherein the physiological signal set includes a plurality of physiological signals of different types;

[0124] The signal processing module 03 is used to extract features from the physiological signal set, and predict the model parameters of the electrical network model according to the signal features of the physiological signal set using a preset prediction model to obtain a prediction model parameter combination;

[0125] The data simulation module 04 is used to apply the prediction model parameter combination to the electrical network model, simulate and generate a corresponding target simulated blood pressure waveform, and determine the blood pressure monitoring result corresponding to the target user based on the target simulated blood pressure waveform.

[0126] Based on the above embodiment, the present invention further provides a terminal, whose principle block diagram can be shown as follows: Figure 7As shown. The terminal includes a processor, a memory, a network interface, and a display screen connected via a system bus. The processor of the terminal is used to provide computing and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the terminal is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a continuous blood pressure monitoring method based on multiple physiological signals and an electrical network is implemented. The display screen of the terminal can be a liquid crystal display or an electronic ink display.

[0127] Those skilled in the art will understand that Figure 7 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the solution of the present invention, and does not constitute a limitation on the terminal to which the solution of the present invention is applied. The specific terminal may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0128] In one implementation, the terminal has one or more programs stored in its memory and is configured to be executed by one or more processors. The one or more programs include instructions for performing a continuous blood pressure monitoring method based on multiple physiological signals and an electrical network.

[0129] Those skilled in the art will appreciate that all or part of the processes in the above-described embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described embodiments. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0130] In summary, the present invention discloses a continuous blood pressure monitoring method based on multiple physiological signals and an electrical network and related equipment. The method obtains an electrical network model pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; wherein the current input corresponding to the electrical network model is determined based on a cardiac cycle time, cardiac systolic ejection time, and cardiac output; the model parameters of the electrical network model include variable heart rate, stroke volume, first resistance value, and capacitance value, and constant cardiac systolic ejection time, second resistance value, and inductance value; obtains a physiological signal set of a target user; wherein the physiological signal set includes several physiological signals of different types; extracts features from the physiological signal set, and predicts the model parameters of the electrical network model based on the signal features of the physiological signal set using a preset prediction model to obtain a prediction model parameter combination; applies the prediction model parameter combination to the electrical network model, and simulates and generates a corresponding target simulated blood pressure waveform, and determines the blood pressure monitoring result corresponding to the target user based on the target simulated blood pressure waveform. The present invention extracts signal features related to blood pressure changes from multiple physiological signals, predicts the model parameters of the electrical network model, and realizes non-invasive continuous blood pressure monitoring based on the relationship between the model parameters and blood pressure changes. Multiple physiological signals can make full use of the advantages and complementary characteristics of different physiological signals to provide more comprehensive information, and the electrical network model can simulate changes in the cardiovascular system, thereby enhancing the accuracy and reliability of continuous blood pressure monitoring.

[0131] It should be understood that the application of the present invention is not limited to the above examples. For those skilled in the art, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A continuous blood pressure monitoring method based on multiple physiological signals and electrical networks, characterized in that: The method comprises: Obtain an electrical network model pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; wherein a current input corresponding to the electrical network model is determined based on a cardiac cycle time, cardiac systolic ejection time, and cardiac output; model parameters of the electrical network model include variable heart rate, stroke volume, first resistance value, and capacitance value, and constant cardiac systolic ejection time, second resistance value, and inductance value; the first resistor and the inductor are connected in parallel to form a first parallel module; the second resistor and the capacitor are connected in parallel to form a second parallel module; and the first parallel module and the second parallel module are connected in series; Acquire a physiological signal set of a target user; wherein the physiological signal set includes a plurality of physiological signals of different types; Extracting features from the physiological signal set includes: dividing each physiological signal in the physiological signal set according to cardiac cycles to obtain signal segments corresponding to each physiological signal; dividing each signal segment into multiple rounds for feature extraction, wherein each signal segment in the same round corresponds to the same cardiac cycle; And the model parameters of the electrical network model are predicted according to the signal characteristics of the physiological signal set through a preset prediction model to obtain a prediction model parameter combination; the physiological signal set includes ECG signals, ICG signals and PPG signals; the prediction model is used to: predict the heart rate of the electrical network model according to the signal characteristics of the ECG signal; predict the output per beat of the electrical network model according to the signal characteristics of the ICG signal; predict the first resistance value and capacitance value of the electrical network model according to the signal characteristics of the PPG signal; the prediction model is optimized in advance through training data, and the method for generating training data includes: pre-determining the numerical values ​​corresponding to each constant model parameter, and setting the parameter grid interval for each variable model parameter to traverse and combine to obtain a number of model parameter combinations; according to each of the model The electrical network model is simulated using a parameter combination to obtain a simulated blood pressure waveform corresponding to each of the model parameter combinations; a plurality of real physiological signal sets and a real blood pressure waveform corresponding to each of the real physiological signal sets are obtained; parameter matching is performed on each of the real blood pressure waveforms and each of the simulated blood pressure waveforms, and a corresponding relationship between each of the real physiological signal sets and each of the model parameter combinations is determined based on the matching result; a plurality of training data are generated based on each of the real physiological signal sets and each of the model parameter combinations, wherein each of the training data includes a pair of the real physiological signal sets and the model parameter combination with a corresponding relationship, the real physiological signal sets are input into the prediction model as training samples, and the model parameter combination is used as a target value corresponding to the training sample to evaluate the accuracy of the predicted value output by the prediction model; The prediction model parameter combination is applied to the electrical network model, and a corresponding target simulated blood pressure waveform is simulated to generate a corresponding blood pressure monitoring result of the target user is determined based on the target simulated blood pressure waveform.

2. The continuous blood pressure monitoring method based on multiple physiological signals and electrical networks according to claim 1, characterized in that: After obtaining the target user's physiological signal set, the following steps are also included: Each physiological signal in the physiological signal set is preprocessed to remove noise; wherein the preprocessing includes: processing low-frequency interference in each physiological signal through wavelet transform, and processing high-frequency interference in each physiological signal through a filter of a preset frequency.

3. The continuous blood pressure monitoring method based on multiple physiological signals and electrical networks according to claim 1, characterized in that: The prediction model is constructed using a gradient-boosted decision tree, and the model optimization method of the prediction model includes: For each round of the multiple rounds of iterative optimization, a training sample of the round is obtained, and a prediction value is generated according to the training sample through all decision trees of the prediction model; Calculate the model residual corresponding to the round according to the predicted value and the corresponding target value; Determining whether the model residual converges to the optimization target; if the model residual does not converge to the optimization target, constructing a new decision tree based on the model residual, and determining the splitting points of the sub-leaves of the decision tree according to a greedy algorithm; Obtain the next round of training samples, and continue to execute the step of generating prediction values ​​according to the training samples through all decision trees of the prediction model until the model residual converges to the optimization target.

4. A continuous blood pressure monitoring system based on multiple physiological signals and electrical networks, characterized in that: The system comprises: An electrical network model is pre-constructed based on a first resistor, a second resistor, a capacitor, and an inductor; the current input corresponding to the electrical network model is determined based on a cardiac cycle time, cardiac systolic ejection time, and cardiac output; the model parameters of the electrical network model include variable heart rate, stroke volume, first resistance value, and capacitance value, and constant cardiac systolic ejection time, second resistance value, and inductance value; the first resistor and the inductor are connected in parallel to form a first parallel module; the second resistor and the capacitor are connected in parallel to form a second parallel module; and the first parallel module and the second parallel module are connected in series; A signal acquisition module, configured to acquire a physiological signal set of a target user; wherein the physiological signal set includes a plurality of physiological signals of different types; a signal processing module configured to perform feature extraction on the physiological signal set, comprising: dividing each physiological signal in the physiological signal set according to cardiac cycles to obtain signal segments corresponding to each physiological signal; and performing feature extraction on each signal segment in multiple rounds, wherein each signal segment in the same round corresponds to the same cardiac cycle; And the model parameters of the electrical network model are predicted according to the signal characteristics of the physiological signal set through a preset prediction model to obtain a prediction model parameter combination; the physiological signal set includes ECG signals, ICG signals and PPG signals; the prediction model is used to: predict the heart rate of the electrical network model according to the signal characteristics of the ECG signal; predict the output per beat of the electrical network model according to the signal characteristics of the ICG signal; predict the first resistance value and capacitance value of the electrical network model according to the signal characteristics of the PPG signal; the prediction model is optimized in advance through training data, and the method for generating training data includes: pre-determining the numerical values ​​corresponding to each constant model parameter, and setting the parameter grid interval for each variable model parameter to traverse and combine to obtain a number of model parameter combinations; according to each of the model The electrical network model is simulated using a parameter combination to obtain a simulated blood pressure waveform corresponding to each of the model parameter combinations; a plurality of real physiological signal sets and a real blood pressure waveform corresponding to each of the real physiological signal sets are obtained; parameter matching is performed on each of the real blood pressure waveforms and each of the simulated blood pressure waveforms, and a corresponding relationship between each of the real physiological signal sets and each of the model parameter combinations is determined based on the matching result; a plurality of training data are generated based on each of the real physiological signal sets and each of the model parameter combinations, wherein each of the training data includes a pair of the real physiological signal sets and the model parameter combination with a corresponding relationship, the real physiological signal sets are input into the prediction model as training samples, and the model parameter combination is used as a target value corresponding to the training sample to evaluate the accuracy of the predicted value output by the prediction model; The data simulation module is used to apply the prediction model parameter combination to the electrical network model, and simulate and generate a corresponding target simulated blood pressure waveform, and determine the blood pressure monitoring result corresponding to the target user based on the target simulated blood pressure waveform.

5. A terminal, characterized in that: The terminal includes a memory and one or more processors; the memory stores one or more programs; the program contains instructions for executing the continuous blood pressure monitoring method based on multiple physiological signals and electrical networks as described in any one of claims 1-3; the processor is used to execute the program.

6. A computer-readable storage medium having a plurality of instructions stored thereon, characterized in that: The instructions are suitable for being loaded and executed by a processor to implement the steps of the continuous blood pressure monitoring method based on multiple physiological signals and electrical networks as described in any one of claims 1 to 3 above.

Images