Noninvasive blood pressure estimation method, device, equipment, storage medium and product

Abstract

The invention provides a non-invasive blood pressure estimation method, device and equipment, a storage medium and a product, and belongs to the technical field of biomedicine, and the method comprises the steps that an ECG signal and an ICG signal are preprocessed; generating physiological feature indexes, wherein the physiological feature indexes comprise conventional indexes and special indexes; determining an incidence relation between the conventional index and the basic peripheral vascular resistance parameter as well as the basic artery compliance parameter; inputting the conventional indexes into a pre-trained machine learning model, and outputting basic peripheral vascular resistance prediction parameters and basic artery compliance prediction parameters; performing dynamic correction according to the special indexes, and generating a target peripheral vascular resistance parameter and a target artery compliance parameter; and calculating and generating a blood pressure estimation result. According to the non-invasive blood pressure estimation method, device and equipment, the storage medium and the product, the accuracy, dynamic nature and stability of non-invasive blood pressure estimation can be improved.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical technology, and in particular relates to a non-invasive blood pressure estimation method, device, equipment, storage medium and product. Background Art

[0002] Blood pressure monitoring is a core method for assessing cardiovascular health. Regular blood pressure measurements can provide early warning of hypertension and various cardiovascular and cerebrovascular diseases. For patients diagnosed with cardiovascular disease, blood pressure monitoring can also assist in assessing the effectiveness of treatment plans, helping to maintain blood pressure within the target range over the long term, thereby significantly reducing the risk of serious complications such as myocardial infarction, stroke, and kidney damage. Therefore, accurate and timely blood pressure monitoring is an indispensable part of clinical diagnosis, health management, and disease prevention.

[0003] In existing technologies, blood pressure monitoring can be broadly categorized into invasive and non-invasive methods. Non-invasive monitoring has been widely used due to its ease of use and low risk. Typically, non-invasive blood pressure monitoring utilizes blood pressure estimation methods based on PPG (pulse wave) and ECG (electrocardiogram) signals. However, the accuracy of PPG signal acquisition is easily affected, leading to errors in the blood pressure estimation results. Furthermore, blood pressure estimation based on PPG and ECG signals fails to adequately reflect the correlation between blood flow dynamics, neural regulation, and vascular characteristics. Consequently, in complex situations, the dynamics and stability of the blood pressure estimation results can be compromised. Summary of the Invention

[0004] In view of this, the present invention aims to propose a non-invasive blood pressure estimation method, device, equipment, storage medium and product to solve the above technical problems.

[0005] To achieve the above object, the technical solution created by the present invention is implemented as follows: In a first aspect, an embodiment of the present invention provides a non-invasive blood pressure estimation method, comprising: Acquire ECG signals and ICG signals, and preprocess the ECG signals and ICG signals; Generate physiological characteristic indicators according to the preprocessed ECG signal and the preprocessed ICG signal, wherein the physiological characteristic indicators include conventional indicators and special indicators; Based on the two-component Windkessel model, the correlation between conventional indices and basal peripheral vascular resistance parameters and basal arterial compliance parameters was determined; Inputting conventional indicators into a pre-trained machine learning model, the machine learning model being configured to: output a basal peripheral vascular resistance prediction parameter and a basal arterial compliance prediction parameter based on the input conventional indicators and according to a correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter; Dynamically correct the basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters according to special indicators to generate target peripheral vascular resistance parameters and target arterial compliance parameters; The target peripheral vascular resistance parameter and the target arterial compliance parameter are input into a two-element Windkessel model, and a blood pressure estimation result is calculated and generated based on conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter.

[0006] Furthermore, the preprocessing of the ECG signal and the ICG signal includes: The ICG signal is pre-processed with adaptive denoising based on wavelet transform; The ECG signal is pre-processed with a 4th-order band-pass FIR filter to remove noise.

[0007] Furthermore, generating physiological characteristic indicators based on the preprocessed ECG signal and the preprocessed ICG signal includes: Based on the improved Pan-Tompkins algorithm based on Shannon energy envelope estimation, R wave detection is performed on the preprocessed ECG signal, and the signal characteristics of the R wave in the ECG signal are extracted; Perform time-frequency analysis on the pre-processed ICG signal based on S-transform, and extract the signal features of points B, C, and X in the ICG signal; Identify the signal characteristics of the R wave in the ECG signal and the signal characteristics of points B, C, and X in the ICG signal to generate physiological characteristic indicators.

[0008] Furthermore, the conventional indicators include: ventricular instantaneous ejection flow, cardiac output, systolic duration, heart rate, end-diastolic initial pressure and cardiac cycle duration; The correlation between the conventional index and the basic peripheral vascular resistance parameter and the basic arterial compliance parameter is expressed as follows: ; In the above formula, is the basic peripheral vascular resistance parameter, is the basic arterial compliance parameter, is the instantaneous ventricular ejection flow, It is the preset value of instantaneous arterial pressure. is a time variable that characterizes moments within the cardiac cycle; The calculation and generation of the blood pressure estimation result based on the conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter is implemented using the following formula: During systole, assuming that the ventricular ejection flow is constant, we have: ; ; During diastole, ventricular ejection flow is set to 0, then: ; In the above formula, is the ventricular ejection flow during systole, is cardiac output, is the duration of the systolic phase, is the heart rate, is systolic blood pressure, is the target peripheral vascular resistance parameter, is the target arterial compliance parameter, is the initial end-diastolic pressure, is diastolic blood pressure, is the cardiac cycle length, is a time variable that represents a moment in the cardiac cycle.

[0009] Furthermore, the special indicators include: standard deviation of cardiac cycle, root mean square of difference between adjacent cardiac cycles, amplitude of point C, slope from point B to point C, area enclosed by point B, point C, and point X, and time difference from point B to point C; The dynamic correction of the basic peripheral vascular resistance prediction parameter and the basic arterial compliance prediction parameter according to the special index to generate the target peripheral vascular resistance parameter and the target arterial compliance parameter is achieved by the following formula: ; ; In the above formula, norm() is the normalization function, is the target peripheral vascular resistance parameter, As the basic peripheral vascular resistance prediction parameter, is the root mean square of the difference between adjacent cardiac cycles, is the standard deviation of the cardiac cycle, is the amplitude at point C, is the target arterial compliance parameter, is the basic arterial compliance prediction parameter, is the slope from point B to point C, is the area enclosed by points B, C, and X. is the time difference from point B to point C, 、 、 、 、 、 and are weight coefficients, where Greater than , Greater than ,and and All less than .

[0010] Furthermore, after the blood pressure estimation result is calculated and generated, the non-invasive blood pressure estimation method further includes: Obtain historical physiological characteristic indicators and historical blood pressure estimation results of multiple consecutive historical cardiac cycles, and calculate changes in historical physiological characteristic indicators and historical blood pressure estimation results between adjacent historical cardiac cycles; Obtaining physiological characteristic indicators of the previous cardiac cycle and calculating the real-time change in physiological characteristic indicators between the current cardiac cycle and the previous cardiac cycle; Inputting the real-time physiological characteristic indicator change into a pre-trained regression model, wherein the regression model is configured to: output a predicted blood pressure estimation result change based on the input real-time physiological characteristic indicator change and a mapping relationship between historical physiological characteristic indicator changes and historical blood pressure estimation result changes; The blood pressure estimation result of the previous cardiac cycle is obtained, and the blood pressure estimation result of the current cardiac cycle is corrected according to the blood pressure estimation result of the previous cardiac cycle and the change in the predicted blood pressure estimation result.

[0011] In a second aspect, an embodiment of the present invention further provides a non-invasive blood pressure estimation device, comprising: A signal acquisition module is used to acquire ECG signals and ICG signals and pre-process the ECG signals and ICG signals; An index generation module is used to generate physiological characteristic indices according to the preprocessed ECG signal and the preprocessed ICG signal, and the physiological characteristic indices include conventional indices and special indices; A determination module, for determining the correlation between conventional indicators and basal peripheral vascular resistance parameters and basal arterial compliance parameters based on a two-element Windkessel model; a prediction module, configured to input conventional indicators into a pre-trained machine learning model, wherein the machine learning model is configured to: output a basal peripheral vascular resistance prediction parameter and a basal arterial compliance prediction parameter based on the input conventional indicators and the correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter; A correction module, used for dynamically correcting the basic peripheral vascular resistance prediction parameter and the basic arterial compliance prediction parameter according to special indicators to generate a target peripheral vascular resistance parameter and a target arterial compliance parameter; The output module is used to input the target peripheral vascular resistance parameter and the target arterial compliance parameter into the two-element Windkessel model, and calculate and generate a blood pressure estimation result based on conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter.

[0012] In a third aspect, an embodiment of the present invention further provides a device, including: one or more processors; a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the non-invasive blood pressure estimation method provided in the above embodiment.

[0013] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer-executable instructions, which, when executed by a computer processor, are used to execute the non-invasive blood pressure estimation method provided in the above embodiment.

[0014] In a fifth aspect, an embodiment of the present invention further provides a computer program product, which stores at least one program code, and the at least one program code is used to be executed by a processor to implement the non-invasive blood pressure estimation method provided in the above embodiment.

[0015] Compared with the prior art, the non-invasive blood pressure estimation method, device, equipment, storage medium and product created by the present invention have the following advantages: The present invention creates a non-invasive blood pressure estimation method, device, equipment, storage medium and product that can generate physiological characteristic indicators including conventional indicators and special indicators through preprocessed ECG signals and ICG signals, and can use a machine learning model to predict and generate basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters based on the conventional indicators. Subsequently, the present invention can also dynamically correct the basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters based on special indicators to obtain target peripheral vascular resistance parameters and target arterial compliance parameters, and finally generate blood pressure estimation results based on conventional indicators, target peripheral vascular resistance parameters and target arterial compliance parameters. Compared with the existing technology, the accuracy of blood pressure estimation results can be improved because the acquisition accuracy of ICG signals is not easily affected. In addition, because the present invention can dynamically correct the basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters, it can fully reflect the correlation between blood flow dynamics, neural regulation and vascular characteristics, so that the blood pressure estimation results still have good dynamics and stability under complex conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which constitute part of the present invention, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings: Figure 1 A flowchart of the non-invasive blood pressure estimation method according to Example 1 of the present invention is provided; Figure 2 This is a schematic structural diagram of the non-invasive blood pressure estimation device according to the second embodiment of the present invention; Figure 3 A schematic diagram of the structure of the device according to the third embodiment of the present invention; Figure 4 A systolic blood pressure scatter plot generated during an actual test using the non-invasive blood pressure estimation method described in the present invention; Figure 5 A diastolic pressure scatter plot generated during an actual test using the non-invasive blood pressure estimation method described in the present invention; Figure 6 A scatter plot of systolic blood pressure error consistency generated during actual testing of the non-invasive blood pressure estimation method created by the present invention; Figure 7 A scatter plot of the consistency of diastolic pressure errors generated during actual testing of the non-invasive blood pressure estimation method created by the present invention; Figure 8 A systolic blood pressure error histogram generated during an actual test of the non-invasive blood pressure estimation method described in the present invention; Figure 9The present invention creates a diastolic pressure error histogram generated during the actual testing process of the non-invasive blood pressure estimation method described in the present invention. DETAILED DESCRIPTION

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and examples. It will 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. It should also be noted that, for ease of description, the accompanying drawings only illustrate portions relevant to the present invention, not all structures.

[0018] Example 1 Figure 1 This is a flow chart of the non-invasive blood pressure estimation method provided in Example 1 of the present invention. Figure 1 As shown, the non-invasive blood pressure estimation method specifically includes the following steps: Step 110: Acquire ECG signals and ICG signals, and pre-process the ECG signals and ICG signals.

[0019] An ECG signal reflects the electrical activity of the heart, while an ICG signal reflects the changes in chest impedance with each heartbeat. In practical applications, both ECG and ICG signals are commonly used medical testing signals. In this embodiment, the ECG and ICG signals are combined to achieve non-invasive blood pressure estimation of the subject.

[0020] To improve the data accuracy in the subsequent processing process, after obtaining the ECG signal and the ICG signal, the ECG signal and the ICG signal should be preprocessed to remove the noise in the signal. In this embodiment, the preprocessing of the ECG signal and the ICG signal may include the following steps: The ICG signal is pre-processed with adaptive denoising based on wavelet transform; The ECG signal is pre-processed with a 4th-order band-pass FIR filter to remove noise.

[0021] Specifically, the adaptive wavelet-based denoising preprocessing of ICG signals described in this embodiment can utilize the Mexican Hat wavelet as the basis function. Due to its excellent time-frequency localization, the Mexican Hat wavelet facilitates capturing sudden changes in signal features such as points B, C, and X in the ICG signal during subsequent processing. Furthermore, the ICG signal should be subjected to a five-scale continuous wavelet decomposition, and different processing strategies should be employed for signals at different scales to facilitate separation of noise from valid signals.

[0022] Scales 1 and 2 primarily contain high-frequency myoelectric noise, which manifests as rapid, irregular fluctuations and significantly interferes with subsequent signal feature extraction. Therefore, hard thresholding is employed: a threshold of 0.8 times the standard deviation of the wavelet coefficients at that scale is set, directly filtering out coefficients with amplitudes less than or equal to the threshold, retaining only high-amplitude coefficients that may contain valid signal edge information.

[0023] For scales 3 and 4, since these signals contain 50Hz power frequency interference and its harmonics, and easily overlap with the effective signal spectrum in the ICG signal, an improved threshold function is used to process them to balance noise suppression and signal fidelity. The expression of the improved threshold function is as follows: ; in, is the ICG signal after wavelet decomposition The scale of wavelet coefficients, After threshold processing The scale of wavelet coefficients, represents the sign function used to preserve the sign of the original wavelet coefficients, The adaptive threshold is formed by traversing all coefficients of the current scale and taking the median of the maximum and minimum values.

[0024] Since the improved threshold function can avoid the discontinuity of the hard threshold at the threshold point, the signal oscillation and the amplitude deviation of the soft threshold can be corrected, thereby ensuring that the processed signal is consistent with the original signal amplitude.

[0025] For scale 5, since this part of the signal corresponds to the effective signal main frequency of the ICG signal (i.e., 30-45 Hz), all coefficients are retained to completely preserve the waveform characteristics of the ICG signal.

[0026] After completing the above processing, the residual noise components of scales 1 and 2 after threshold processing should be eliminated, and the processed coefficients of scales 3, 4, and 5 should be integrated for reconstruction. Ultimately, the baseline drift, 50 Hz power frequency interference, and high-frequency electromyographic noise can be effectively suppressed, completing the denoising preprocessing of the ICG signal.

[0027] In this embodiment, the ECG signal is subjected to denoising preprocessing based on a 4th-order bandpass FIR filter, and the filter specifications are as follows: Order: 4th order (reducing computational complexity while ensuring filtering effect); Passband range: 5-30Hz (covering the main energy range of the QRS wave to ensure the integrity of the QRS wave waveform); Stopband design: less than 0.5Hz (attenuation ≥60dB, suppressing low-frequency interference such as P waves and T waves) and greater than 40Hz (attenuation ≥60dB, suppressing high-frequency myoelectric noise); Window function: Hanning window is used with a transition band width less than 2 Hz to reduce the distortion effect of passband ripple on the signal.

[0028] The above filter can reduce the interference of P wave and T wave in ECG waveform on QRS wave extraction process and the interference of noise in ICG waveform, thereby improving signal clarity and accuracy of feature extraction.

[0029] Step 120: Generate physiological characteristic indicators based on the preprocessed ECG signal and the preprocessed ICG signal, and the physiological characteristic indicators include conventional indicators and special indicators.

[0030] After the ECG signal and ICG signal are preprocessed, this embodiment will generate physiological characteristic indicators including conventional indicators and special indicators based on the preprocessed ECG signal and the preprocessed ICG signal, so as to obtain blood pressure estimation results with good accuracy, dynamics and stability through conventional indicators and special indicators in the subsequent processing process.

[0031] Optionally, generating physiological characteristic indicators based on the preprocessed ECG signal and the preprocessed ICG signal can be specifically optimized into the following steps: Based on the improved Pan-Tompkins algorithm based on Shannon energy envelope estimation, R wave detection is performed on the preprocessed ECG signal, and the signal characteristics of the R wave in the ECG signal are extracted; Perform time-frequency analysis on the pre-processed ICG signal based on S-transform, and extract the signal features of points B, C, and X in the ICG signal; Identify the signal characteristics of the R wave in the ECG signal and the signal characteristics of points B, C, and X in the ICG signal to generate physiological characteristic indicators.

[0032] Among them, the Pan-Tompkins algorithm improved based on Shannon energy envelope estimation performs R wave detection on the preprocessed ECG signal, which can improve the robustness of identifying R waves of different amplitudes in complex noise environments. The specific steps include: First, the slope of the preprocessed ECG signal is focused on based on the classic Pan-Tompkins algorithm, and a first-order forward difference transform is performed. Its mathematical expression is as follows: ; In the above formula, is the preprocessed ECG signal sampling value, is the difference result.

[0033] The above processing can enhance the steep edge characteristics of the QRS complex, further suppress the interference of the slowly changing T wave and P wave, output a bipolar differential signal, and provide the original input for subsequent energy feature extraction.

[0034] Subsequently, the Shannon energy envelope is extracted from the processed signal, and its mathematical expression is as follows: ; In the above formula, is the instantaneous energy sequence, is the minimum value (usually 10 -8 ), used to avoid logarithmic operation exceptions, To normalize the difference results so that their values ​​fall within The threshold is kept within the range of 0 to 1000, which can eliminate amplitude differences (due to individual differences, lead differences, or electrode contact changes) and make the threshold more stable.

[0035] Through the above nonlinear transformation, the significant features of the QRS complex are selectively amplified to form an instantaneous energy sequence ( ), followed by a zero-phase low-pass filter (with a cutoff frequency set to 0.5Hz), resulting in a smoothed Shannon Energy Envelope (SEE), whose local peaks can be precisely mapped to the QRS complex position. The zero-phase filter eliminates group delay / phase lag, preventing the peak position from being postponed, thus improving the temporal accuracy of R-peak location. A dynamic threshold is then set based on statistical methods, combined with a local peak search strategy. Each time a candidate peak is found, a physiological refractory period (e.g., 200-250ms) is set to suppress false detections (preventing the same QRS rising and falling edges / high-frequency ripples from being counted as multiple peaks). The peak is initially located on the envelope, and a refined search is then performed near this peak on the original / bandpassed ECG signal (e.g., the point of maximum rising slope or the point of local maximum amplitude) to accurately extract the R-peak position. Ultimately, the signal characteristics of the R wave in the ECG signal are extracted based on the R-peak.

[0036] Furthermore, based on the detected R-peak positions, the signal segment between two adjacent R-peaks can be defined as a complete cardiac cycle, completing cardiac cycle division. Accordingly, after completing cardiac cycle division of the ECG signal, the ICG signal can be further divided into cycles based on the same time intervals, ensuring that subsequent processing is performed within the same cardiac cycle.

[0037] In this embodiment, time-frequency analysis of the preprocessed ICG signal using the S transform can enhance the local waveform variation characteristics of key feature points in the ICG signal. Compared with the traditional wavelet transform, which is a time-scale analysis and therefore has a non-proportional scale to frequency and a complex transformation process, the result is susceptible to noise. The S transform better adapts to the characteristics of the ICG signal, effectively enhancing the local waveform variation characteristics corresponding to the key feature points. Therefore, the S transform can extract the signal features of points B, C, and X in the ICG signal with greater accuracy.

[0038] Specifically, since the key feature points in the ICG signal mainly include point B (the starting point of left ventricular ejection), point C (the peak point of ventricular contraction), and point X (the end point of left ventricular ejection), this embodiment will illustrate the specific processing method and effect of the S transform with respect to the above key feature points.

[0039] When processing point B, the S-transform is used to capture the mutation characteristics of the ICG signal at the initial stage, and the local waveform change corresponding to this point is enhanced to make its starting mark clearer.

[0040] When processing point C, the sensitivity of the S transform to the signal extreme value characteristics is utilized to highlight the impedance change peak corresponding to the maximum ventricular contraction state in the ICG signal, and to enhance the energy aggregation characteristics of point C in the time-frequency domain, thereby improving its detection accuracy as a key indicator for evaluating cardiac pump function.

[0041] When processing point X, the S-transform is used to identify the characteristics of the termination of the ICG signal trend, so as to clearly define this point as the sign of the end of ejection.

[0042] The above processing can significantly enhance the features of the ICG signal, such as the initial mutation at point B, the peak extreme value at point C, and the trend termination at point X, thereby facilitating the extraction of signal features at points B, C, and X.

[0043] After extracting the signal characteristics of the R wave in the ECG signal and the signal characteristics of points B, C, and X in the ICG signal, this embodiment generates physiological characteristic indicators based on these signal characteristics. It should be noted that the process of generating physiological characteristic indicators based on the signal characteristics of the R wave in the ECG signal and the signal characteristics of points B, C, and X in the ICG signal is a mature existing technology. In practical applications, personnel can analyze these signal characteristics in the time domain, frequency domain, amplitude, and waveform morphology to generate corresponding physiological characteristic indicators. Alternatively, a pre-trained neural network model can be used instead of manual analysis to convert signal characteristics into physiological characteristic indicators.

[0044] Step 130: Based on the two-element Windkessel model, determine the correlation between the conventional index and the basic peripheral vascular resistance parameter and the basic arterial compliance parameter.

[0045] The two-element Windkessel model is a simplified model used to simulate cardiovascular hemodynamics. It approximates the arterial system as a second-order circuit consisting of resistance and capacitance, where resistance represents the resistance of the vessel wall to blood flow, and capacitance reflects the elastic compliance of the arterial wall. By describing the dynamic relationship between pressure and flow, this model effectively simulates the propagation and attenuation of pulse waves in the arterial system. Therefore, this example uses the two-element Windkessel model to determine the correlation between conventional indicators and the basal peripheral vascular resistance and basal arterial compliance parameters, facilitating the prediction of these parameters in subsequent processing.

[0046] Specifically, to predict the basal peripheral vascular resistance parameter and the basal arterial compliance parameter, the conventional indices in this embodiment may include ventricular instantaneous ejection flow. Accordingly, the correlation between the conventional indices and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter is expressed as follows: ; In the above formula, is the basic peripheral vascular resistance parameter, is the basic arterial compliance parameter, is the instantaneous ventricular ejection flow, It is the preset value of instantaneous arterial pressure. is a time variable that represents a moment in the cardiac cycle.

[0047] It should be noted that the preset instantaneous arterial pressure values ​​described in this embodiment are based on the normal instantaneous arterial pressure of healthy individuals. When the time variable indicates the current systolic phase, the preset instantaneous arterial pressure value can be set between 90 and 139 (corresponding to normal human systolic pressure). When the time variable indicates the current diastolic phase, the preset instantaneous arterial pressure value can be set between 60 and 89 (corresponding to normal human diastolic pressure). Accordingly, when setting the instantaneous arterial pressure value within the aforementioned ranges, staff can also adjust the value based on the subject's actual age, so that the larger the subject's actual age, the larger the preset instantaneous arterial pressure value.

[0048] For example, when determining the conventional ventricular instantaneous ejection flow (ITF), the ventricular ejection time can be determined based on the distance from point B to point X, and the peak of the ventricular rapid ejection period can be determined by the amplitude at point C. The stroke volume is then calculated using the common Kubicek formula, and the corresponding flow waveform is generated using the ICG signal to ultimately obtain the ventricular instantaneous ejection flow. In the prior art, the specific method for determining the ventricular instantaneous ejection flow based on the signal characteristics of the ICG signal is well known to those skilled in the art and does not constitute the core invention of this application. Therefore, the specific calculation steps are not detailed in this application.

[0049] Step 140: Input the conventional indicators into a pre-trained machine learning model, wherein the machine learning model is configured to: based on the input conventional indicators, output the basal peripheral vascular resistance prediction parameter and the basal arterial compliance prediction parameter according to the correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter.

[0050] In this embodiment, a common random forest model among existing machine learning models can be used to predict the basal peripheral vascular resistance prediction parameter and the basal arterial compliance prediction parameter. During use, after conventional indicators are input into the random forest model, the random forest model predicts the basal peripheral vascular resistance prediction parameter and the basal arterial compliance prediction parameter based on the correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter determined in step 130. The model then outputs the basal peripheral vascular resistance prediction parameter and the basal arterial compliance prediction parameter as the results for use in subsequent steps.

[0051] As an optional implementation of this embodiment, to improve the accuracy of the machine learning model's prediction results, the physiological characteristic indicators may also include other multidimensional indicators. When training the machine learning model using historical data, these other historical multidimensional indicators can be input into the model simultaneously with historical conventional indicators. This allows the model to learn the mapping relationship between these other historical multidimensional indicators and historical prediction results (i.e., historical basal peripheral vascular resistance prediction parameters and historical basal arterial compliance prediction parameters generated based on historical conventional indicators) during training. Therefore, in practical applications, by simultaneously inputting conventional indicators and other multidimensional indicators into the machine learning model, more accurate basal peripheral vascular resistance prediction parameters and basal arterial compliance prediction parameters can be obtained.

[0052] By way of example and not limitation, other multidimensional indicators include, but are not limited to, the following: the time difference between point R and point B, the time difference between point R and point C, the time difference between point R and point X, the time difference between point C and point X, the time of the right intersection of the 75% horizontal line, the time of the left intersection of the 75% horizontal line, the time difference between the left and right intersections of the 75% horizontal line, the time of the right intersection of the 50% horizontal line, the time of the left intersection of the 50% horizontal line, the time difference between the left and right intersections of the 50% horizontal line, the time of the right intersection of the 25% horizontal line, the time of the left intersection of the 25% horizontal line, the time difference between the left and right intersections of the 25% horizontal line, the time when the maximum derivative occurs, the time when the minimum derivative occurs, the time difference between the maximum derivative and the minimum derivative, the amplitude of point B, the amplitude of point X, the amplitude difference between point C and point B, the amplitude difference between point C and point X, the original signal amplitude corresponding to the maximum derivative, the original signal amplitude corresponding to the minimum derivative, the maximum derivative amplitude between B and C value, minimum derivative amplitude between C and X, ratio of amplitude at point B to amplitude at point C, ratio of amplitude at point B to amplitude at point X, ratio of amplitude at point C to amplitude at point X, ratio of amplitude at point C to amplitude at point B, ratio of amplitude at point X to amplitude at point B, ratio of amplitude at point X to amplitude at point C, ratio of time difference between 75% horizontal line and cardiac cycle time, ratio of time difference between 50% horizontal line and cardiac cycle time, ratio of time difference between 25% horizontal line and cardiac cycle time, time from B to C The following data were collected: the ratio of the time difference between cardiac intervals and the cardiac cycle time, the ratio of the time difference between C and X to the cardiac cycle time, the ratio of the time difference between B and X to the cardiac cycle time, the ratio of the systolic time to the diastolic time, the slope of C to X, the mean cardiac interval, the number of times the interval difference between adjacent cardiac cycles exceeded 50 milliseconds, the percentage of NN50 in the total number of RR intervals, the median cardiac cycle, the minimum cardiac cycle value, the maximum cardiac cycle value, and the difference between the maximum and minimum cardiac cycle values.

[0053] Step 150: Dynamically modify the basal peripheral vascular resistance prediction parameter and the basal arterial compliance prediction parameter according to the special index to generate a target peripheral vascular resistance parameter and a target arterial compliance parameter.

[0054] After outputting the basal peripheral vascular resistance prediction parameters and basal arterial compliance prediction parameters through the machine learning model, this embodiment will also dynamically correct them through special indicators to generate target peripheral vascular resistance parameters and target arterial compliance parameters, so that the target peripheral vascular resistance parameters and target arterial compliance parameters can accurately reflect the impact of neural regulation on peripheral vascular resistance and the impact of blood flow dynamics on arterial compliance, thereby making subsequent blood pressure estimation results more accurate.

[0055] Specifically, the special indicators in this embodiment may include: the standard deviation of the cardiac cycle, the root mean square of the difference between adjacent cardiac cycles, the amplitude of point C, the slope from point B to point C, the area enclosed by points B, C, and X, and the time difference from point B to point C.

[0056] Accordingly, the basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters are dynamically modified according to special indicators to generate target peripheral vascular resistance parameters and target arterial compliance parameters, which are achieved using the following formula: ; ; In the above formula, norm() is the normalization function, is the target peripheral vascular resistance parameter, As the basic peripheral vascular resistance prediction parameter, is the root mean square of the difference between adjacent cardiac cycles, is the standard deviation of the cardiac cycle, is the amplitude at point C, is the target arterial compliance parameter, is the basic arterial compliance prediction parameter, is the slope from point B to point C, is the area enclosed by points B, C, and X. is the time difference from point B to point C, 、 、 、 、 、 and are weight coefficients, where Greater than , Greater than ,and and All less than .

[0057] Since peripheral vascular resistance is affected by vascular tension, neural regulation, and blood flow intensity, this embodiment dynamically corrects the basic peripheral vascular resistance prediction parameters using the root mean square of the difference between adjacent cardiac cycles, the standard deviation of the cardiac cycle, and the amplitude of point C to obtain the target peripheral vascular resistance parameters.

[0058] Specifically, the root mean square of the difference between adjacent cardiac cycles can reflect the activity of the vagus nerve. When the root mean square of the difference between adjacent cardiac cycles is high, it proves that the vagus nerve is active, which can promote vasodilation. At this time, peripheral vascular resistance should decrease.

[0059] The standard deviation of the cardiac cycle can reflect the overall heart rate variability. When the standard deviation of the cardiac cycle is high, it indicates that unstable neural regulation (such as sympathetic nerve activity) occurs at this time, and peripheral vascular resistance should increase.

[0060] The amplitude of point C can reflect the blood flow intensity during ventricular contraction. When the amplitude of point C is high, it indicates that the current blood flow intensity is high, and usually the peripheral vascular resistance should decrease.

[0061] When generating the target peripheral vascular resistance parameter, this embodiment also introduces 、 and As a weight coefficient, it reflects the contribution of each special index to the target peripheral vascular resistance parameter. Because the influence of vagus nerve activity on peripheral vascular resistance is greater than that of overall heart rate variability, Should be greater than .

[0062] Because arterial compliance is affected by blood flow dynamics (i.e., ejection velocity and cumulative effect) and neural regulation, this embodiment dynamically corrects the baseline arterial compliance prediction parameters using the slope from point B to point C, the area enclosed by points B, C, and X, the time difference from point B to point C, and the root mean square of the difference between adjacent cardiac cycles to obtain the target arterial compliance parameters.

[0063] Specifically, the slope from point B to point C can reflect the blood acceleration rate. When the slope from point B to point C is high, it indicates a rapid blood flow change. At this time, the arterial elasticity is good, so the arterial compliance strain is large.

[0064] The area enclosed by points B, C, and X can reflect the cumulative blood flow effect of the blood mapping process. When the area enclosed by points B, C, and X is larger, it proves that the buffering capacity is stronger at this time, and therefore the arterial compliance strain is large.

[0065] The time difference between point B and point C can reflect the blood ejection time. When the time difference between point B and point C is small, it proves that it is a rapid ejection state, the arterial elasticity is good, and therefore the arterial compliance strain is large.

[0066] The root mean square of the difference between adjacent cardiac cycles can reflect the activity of the vagus nerve. When the root mean square of the difference between adjacent cardiac cycles is high, it proves that the vagus nerve is active, which helps to improve vascular elasticity. At this time, arterial compliance should also increase.

[0067] When generating the target arterial compliance parameter, this embodiment also introduces 、 、 and As a weight coefficient, it reflects the contribution of each special index to the target arterial compliance parameter. Among them, the effects of ejection time and vagus nerve activity on arterial compliance are relatively small, and the effect of ejection acceleration rate on arterial compliance is greater than the cumulative blood flow effect. Therefore, Should be greater than ,and and Should be less than .

[0068] It should be additionally explained that the process of generating the above-mentioned special indicators through the signal characteristics of ECG signals and the signal characteristics of ICG signals are also existing technologies. The specific methods are already well known to those skilled in the art and do not belong to the core invention points of this application, so they will not be repeated in this application.

[0069] Step 160: Input the target peripheral vascular resistance parameter and the target arterial compliance parameter into a two-element Windkessel model, and calculate and generate a blood pressure estimation result based on conventional indicators, the target peripheral vascular resistance parameter, and the target arterial compliance parameter.

[0070] After obtaining the target peripheral vascular resistance parameter and the target arterial compliance parameter, this embodiment will calculate and generate a blood pressure estimation result based on the two-element Windkessel model, conventional indicators, the target peripheral vascular resistance parameter, and the target arterial compliance parameter to achieve the purpose of non-invasive blood pressure estimation for the subject.

[0071] Specifically, the conventional indicators in this embodiment may also include: cardiac output, systolic duration, heart rate, end-diastolic initial pressure and cardiac cycle duration.

[0072] Accordingly, based on conventional indicators, target peripheral vascular resistance parameters, and target arterial compliance parameters, blood pressure estimation results are calculated and generated, which can be achieved using the following formula: During systole, assuming that the ventricular ejection flow is constant, we have: ; ; During diastole, ventricular ejection flow is set to 0, then: ; In the above formula, is the ventricular ejection flow during systole, is cardiac output, is the duration of the systolic phase, is the heart rate, is systolic blood pressure, is the target peripheral vascular resistance parameter, is the target arterial compliance parameter, is the initial end-diastolic pressure, is diastolic blood pressure, is the cardiac cycle length, is a time variable that represents a moment in the cardiac cycle.

[0073] It should be additionally explained that the process of generating the above-mentioned conventional indicators through the signal characteristics of ECG signals and the signal characteristics of ICG signals are also existing technologies. The specific methods are already well known to those skilled in the art and do not belong to the core invention points of this application, so they will not be repeated in this application.

[0074] By calculating with the above formula, the estimated blood pressure results (i.e., diastolic and systolic pressures) of the subject in one cardiac cycle can be obtained. The staff can repeat the above steps in each cardiac cycle to complete continuous blood pressure estimation.

[0075] Optionally, after calculating and generating the blood pressure estimation result, the non-invasive blood pressure estimation method in this embodiment further adds the following steps: Obtain historical physiological characteristic indicators and historical blood pressure estimation results of multiple consecutive historical cardiac cycles, and calculate changes in historical physiological characteristic indicators and historical blood pressure estimation results between adjacent historical cardiac cycles; Obtaining physiological characteristic indicators of the previous cardiac cycle and calculating the real-time change in physiological characteristic indicators between the current cardiac cycle and the previous cardiac cycle; Inputting the real-time physiological characteristic indicator change into a pre-trained regression model, wherein the regression model is configured to: output a predicted blood pressure estimation result change based on the input real-time physiological characteristic indicator change and a mapping relationship between historical physiological characteristic indicator changes and historical blood pressure estimation result changes; The blood pressure estimation result of the previous cardiac cycle is obtained, and the blood pressure estimation result of the current cardiac cycle is corrected according to the blood pressure estimation result of the previous cardiac cycle and the change in the predicted blood pressure estimation result.

[0076] The above process can reduce the cumulative error and systematic deviation of this method in long time series data. The blood pressure estimation result of the current cardiac cycle is corrected by predicting the change in the blood pressure estimation result. Therefore, the estimation stability and anti-drift ability of this method in real continuous data can be effectively improved. It is particularly suitable for error control and result consistency optimization in long-term monitoring tasks.

[0077] This embodiment provides a non-invasive blood pressure estimation method, which can achieve denoising by preprocessing ECG signals and ICG signals, and generate physiological characteristic indicators of conventional indicators and special indicators based on signal characteristics. Subsequently, a machine learning model is used to generate basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters based on conventional indicators, and these are corrected by special indicators to obtain target peripheral vascular resistance parameters and target arterial compliance parameters, so that the target peripheral vascular resistance parameters and target arterial compliance parameters fully reflect the influence of neural regulation and blood flow dynamics. Finally, a blood pressure estimation result is generated through conventional indicators, target peripheral vascular resistance parameters and target arterial compliance parameters. Compared with the existing technology, it can improve the accuracy of blood pressure estimation results and enable the blood pressure estimation results to still have good dynamics and stability under complex conditions.

[0078] In addition, the inventors conducted practical application tests based on the non-invasive blood pressure estimation method provided in this embodiment. After testing, the cumulative error of systolic blood pressure estimation performed by this method was less than 5 mmHg in 84.00%, the cumulative error was less than 10 mmHg in 96.00%, and the cumulative error was less than 15 mmHg in 100.00%. The cumulative error of diastolic blood pressure estimation was less than 5 mmHg in 98.00%, the cumulative error was less than 10 mmHg in 100.00%, and the cumulative error was less than 15 mmHg in 100.00%. According to the British Hypertension Society (BHS) standards, the systolic and diastolic blood pressure estimations of this method are both rated A. The experimental results are shown in the following table:

[0079] in addition, Figure 4 This is the systolic blood pressure scatter plot generated by this method during the actual test process. Figure 5 This is the diastolic pressure scatter plot generated by this method during the actual test process. Figure 6 This is a scatter plot of systolic blood pressure error consistency formed by this method during the actual test process. Figure 7 This is the scatter plot of diastolic pressure error consistency formed by this method in the actual test process. Figure 8 is the systolic blood pressure error histogram formed during the actual test process of this method, Figure 9 This is the diastolic pressure error histogram formed during the actual test process of this method. Figure 4-Figure 9In actual blood pressure estimation tests based on this method, the predicted values ​​for systolic and diastolic blood pressures were highly consistent with the true values. The 95% agreement intervals for systolic and diastolic blood pressures were -0.88±7.15 mmHg and -0.14±3.44 mmHg, respectively, both below the clinically accepted ±10 mmHg threshold. Error histograms showed that approximately 85% and 98% of the errors for systolic and diastolic blood pressures fell within ±5 mmHg, with the diastolic pressure being particularly concentrated. This demonstrates the robustness of this method on individualized datasets and provides reliable support for the clinical application of non-invasive blood pressure monitoring.

[0080] Example 2 Figure 2 FIG. 1 is a schematic diagram of the structure of the non-invasive blood pressure estimation device provided in the second embodiment of the present invention. Figure 2 As shown, the non-invasive blood pressure estimation device includes: A signal acquisition module 210 is used to acquire ECG signals and ICG signals and pre-process the ECG signals and ICG signals; An index generating module 220 is configured to generate physiological characteristic indices based on the preprocessed ECG signal and the preprocessed ICG signal, wherein the physiological characteristic indices include conventional indices and special indices; a determination module 230 for determining, based on a two-element Windkessel model, a correlation between conventional indicators and a basal peripheral vascular resistance parameter and a basal arterial compliance parameter; Prediction module 240 is configured to input conventional indicators into a pre-trained machine learning model, wherein the machine learning model is configured to: output a basal peripheral vascular resistance prediction parameter and a basal arterial compliance prediction parameter based on the input conventional indicators and the correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter; A correction module 250 is used to dynamically correct the basic peripheral vascular resistance prediction parameter and the basic arterial compliance prediction parameter according to special indicators to generate a target peripheral vascular resistance parameter and a target arterial compliance parameter; The output module 260 is used to input the target peripheral vascular resistance parameter and the target arterial compliance parameter into the two-element Windkessel model, and calculate and generate a blood pressure estimation result based on the conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter.

[0081] This embodiment provides a non-invasive blood pressure estimation device that can generate physiological characteristic indicators based on ECG signals and ICG signals through the cooperation of a signal acquisition module and an indicator generation module. At the same time, it can predict and generate basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters through the cooperation of a determination module and a prediction module, and can dynamically correct the basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters through a correction module, thereby forming target peripheral vascular resistance parameters and target arterial compliance parameters that can reflect neural regulation and blood flow dynamics. Finally, a blood pressure estimation result can be generated through an output module, thereby achieving accurate, stable, and highly dynamic non-invasive blood pressure estimation.

[0082] Based on the above embodiment, the signal acquisition module includes: An ICG signal preprocessing unit, used for performing adaptive denoising preprocessing on the ICG signal based on wavelet transform; The ECG signal preprocessing unit is used to perform denoising preprocessing on the ECG signal based on a 4th-order bandpass FIR filter.

[0083] Based on the above embodiment, the indicator generation module includes: An ECG signal feature extraction unit is used to perform R wave detection on the preprocessed ECG signal based on the Pan-Tompkins algorithm improved by Shannon energy envelope estimation, and to extract the signal features of the R wave in the ECG signal; An ICG signal feature extraction unit is used to perform time-frequency analysis on the pre-processed ICG signal based on S-transformation and extract signal features of points B, C, and X in the ICG signal; The generating unit is used to identify the signal characteristics of the R wave in the ECG signal and the signal characteristics of point B, point C and point X in the ICG signal, and generate physiological characteristic indicators.

[0084] Based on the above embodiment, the non-invasive blood pressure estimation device further includes: The correction module is used to obtain historical physiological characteristic indicators and historical blood pressure estimation results of multiple consecutive historical cardiac cycles, calculate the changes in historical physiological characteristic indicators and historical blood pressure estimation results between adjacent historical cardiac cycles; obtain the physiological characteristic indicators of the previous cardiac cycle, calculate the changes in real-time physiological characteristic indicators between the current cardiac cycle and the previous cardiac cycle; input the changes in real-time physiological characteristic indicators into a pre-trained regression model, and the regression model is configured to: based on the input real-time physiological characteristic indicator changes, output the predicted blood pressure estimation result changes according to the mapping relationship between the historical physiological characteristic indicator changes and the historical blood pressure estimation result changes; obtain the blood pressure estimation result of the previous cardiac cycle, and correct the blood pressure estimation result of the current cardiac cycle according to the blood pressure estimation result of the previous cardiac cycle and the predicted blood pressure estimation result changes.

[0085] The non-invasive blood pressure estimation device provided in the embodiment of the present invention can execute the non-invasive blood pressure estimation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0086] Example 3 Figure 3 This is a schematic diagram of the structure of the device provided in Example 3 of the present invention. Figure 3 A block diagram of an exemplary device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 3 The device 12 shown is only an example and should not bring any limitation to the functionality and scope of use of the embodiments of the present invention.

[0087] like Figure 3 As shown, device 12 is implemented as a general-purpose computing device. Components of device 12 may include, but are not limited to, one or more processors or processing units 16, a system memory 28, and a bus 18 that connects various system components (including system memory 28 and processing unit 16).

[0088] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0089] Device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by device 12, including volatile and non-volatile media, removable and non-removable media.

[0090] System memory 28 may include computer system readable media in the form of volatile memory, such as RAM 30 and / or cache 32. Device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be configured to read and write non-removable, non-volatile magnetic media ( Figure 3 Not shown, usually called a "hard drive"). Although Figure 3 Although not shown, a magnetic disk drive for reading and writing to a removable non-volatile magnetic disk (e.g., a "floppy disk"), as well as an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 via one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of various embodiments of the present invention.

[0091] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may include an implementation of a network environment. Program modules 42 generally implement the functions and / or methodologies of the embodiments described herein.

[0092] Device 12 may also communicate with one or more external devices 14 (e.g., a keyboard, pointing device, display 24, etc.), one or more devices that enable a user to interact with device 12, and / or any device that enables device 12 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). Such communication may occur via I / O interface 22. Furthermore, device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via network adapter 20. As shown, network adapter 20 communicates with other modules of device 12 via bus 18. It should be understood that, although not shown, other hardware and / or software modules may be used in conjunction with device 12, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0093] The processing unit 16 executes various functional applications and data processing by running programs stored in the system memory 28, such as implementing the non-invasive blood pressure estimation method provided by the embodiment of the present invention.

[0094] Example 4 Embodiment 4 of the present invention further provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform any of the non-invasive blood pressure estimation methods provided in the above embodiments.

[0095] The computer storage medium of the embodiments of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0096] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0097] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0098] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0099] Example 5 Embodiments of the present invention provide a computer program product, comprising a computer program, and the computer program product includes at least one program code. The computer program may be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the non-invasive blood pressure estimation method provided in the above-described embodiments.

[0100] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A non-invasive blood pressure estimation method, characterized in that include: Acquire ECG signals and ICG signals, and preprocess the ECG signals and ICG signals; Generate physiological characteristic indicators according to the preprocessed ECG signal and the preprocessed ICG signal, wherein the physiological characteristic indicators include conventional indicators and special indicators; Based on the two-component Windkessel model, the correlation between conventional indices and basal peripheral vascular resistance parameters and basal arterial compliance parameters was determined; Inputting conventional indicators into a pre-trained machine learning model, the machine learning model being configured to: output a basal peripheral vascular resistance prediction parameter and a basal arterial compliance prediction parameter based on the input conventional indicators and according to a correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter; Dynamically correct the basic peripheral vascular resistance prediction parameters and basic arterial compliance prediction parameters according to special indicators to generate target peripheral vascular resistance parameters and target arterial compliance parameters; The target peripheral vascular resistance parameter and the target arterial compliance parameter are input into a two-element Windkessel model, and a blood pressure estimation result is calculated and generated based on conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter.

2. The non-invasive blood pressure estimation method according to claim 1, wherein: The preprocessing of the ECG signal and the ICG signal comprises: The ICG signal is pre-processed with adaptive denoising based on wavelet transform; The ECG signal is pre-processed with a 4th-order band-pass FIR filter to remove noise.

3. The non-invasive blood pressure estimation method according to claim 1, wherein: Generating physiological characteristic indicators according to the preprocessed ECG signal and the preprocessed ICG signal includes: Based on the improved Pan-Tompkins algorithm based on Shannon energy envelope estimation, R wave detection is performed on the preprocessed ECG signal, and the signal characteristics of the R wave in the ECG signal are extracted; Perform time-frequency analysis on the pre-processed ICG signal based on S-transform, and extract the signal features of points B, C, and X in the ICG signal; Identify the signal characteristics of the R wave in the ECG signal and the signal characteristics of points B, C, and X in the ICG signal to generate physiological characteristic indicators.

4. The non-invasive blood pressure estimation method according to claim 1, wherein: The conventional indicators include: ventricular instantaneous ejection flow, cardiac output, systolic duration, heart rate, initial end-diastolic pressure and cardiac cycle duration; The correlation between the conventional index and the basic peripheral vascular resistance parameter and the basic arterial compliance parameter is expressed as follows: ； In the above formula, is the basic peripheral vascular resistance parameter, is the basic arterial compliance parameter, is the instantaneous ventricular ejection flow, It is the preset value of instantaneous arterial pressure. is a time variable that characterizes moments within the cardiac cycle; The calculation and generation of the blood pressure estimation result based on the conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter is implemented using the following formula: During systole, assuming that the ventricular ejection flow is constant, we have: ； ； During diastole, ventricular ejection flow is set to 0, then: ； In the above formula, is the ventricular ejection flow during systole, is cardiac output, is the duration of the systolic phase, is the heart rate, is systolic blood pressure, is the target peripheral vascular resistance parameter, is the target arterial compliance parameter, is the initial end-diastolic pressure, is diastolic blood pressure, is the cardiac cycle length, is a time variable that represents a moment in the cardiac cycle.

5. The non-invasive blood pressure estimation method according to claim 1, wherein: The special indicators include: standard deviation of cardiac cycle, root mean square of difference between adjacent cardiac cycles, amplitude of point C, slope from point B to point C, area enclosed by point B, point C and point X, and time difference from point B to point C; The dynamic correction of the basic peripheral vascular resistance prediction parameter and the basic arterial compliance prediction parameter according to the special index to generate the target peripheral vascular resistance parameter and the target arterial compliance parameter is achieved by the following formula: ； ； In the above formula, norm() is the normalization function, is the target peripheral vascular resistance parameter, As the basic peripheral vascular resistance prediction parameter, is the root mean square of the difference between adjacent cardiac cycles, is the standard deviation of the cardiac cycle, is the amplitude at point C, is the target arterial compliance parameter, is the basic arterial compliance prediction parameter, is the slope from point B to point C, is the area enclosed by points B, C, and X. is the time difference from point B to point C, 、 、 、 、 、 and are weight coefficients, where Greater than , Greater than ,and and All less than .

6. The non-invasive blood pressure estimation method according to claim 1, wherein: After calculating and generating the blood pressure estimation result, the non-invasive blood pressure estimation method further includes: Obtain historical physiological characteristic indicators and historical blood pressure estimation results of multiple consecutive historical cardiac cycles, and calculate changes in historical physiological characteristic indicators and historical blood pressure estimation results between adjacent historical cardiac cycles; Obtaining physiological characteristic indicators of the previous cardiac cycle and calculating the real-time change in physiological characteristic indicators between the current cardiac cycle and the previous cardiac cycle; Inputting the real-time physiological characteristic indicator change into a pre-trained regression model, wherein the regression model is configured to: output a predicted blood pressure estimation result change based on the input real-time physiological characteristic indicator change and a mapping relationship between historical physiological characteristic indicator changes and historical blood pressure estimation result changes; The blood pressure estimation result of the previous cardiac cycle is obtained, and the blood pressure estimation result of the current cardiac cycle is corrected according to the blood pressure estimation result of the previous cardiac cycle and the change in the predicted blood pressure estimation result.

7. A non-invasive blood pressure estimation device, characterized in that: include: A signal acquisition module is used to acquire ECG signals and ICG signals and pre-process the ECG signals and ICG signals; An index generation module is used to generate physiological characteristic indices according to the preprocessed ECG signal and the preprocessed ICG signal, and the physiological characteristic indices include conventional indices and special indices; A determination module, for determining the correlation between conventional indicators and basal peripheral vascular resistance parameters and basal arterial compliance parameters based on a two-element Windkessel model; a prediction module, configured to input conventional indicators into a pre-trained machine learning model, wherein the machine learning model is configured to: output a basal peripheral vascular resistance prediction parameter and a basal arterial compliance prediction parameter based on the input conventional indicators and the correlation between the conventional indicators and the basal peripheral vascular resistance parameter and the basal arterial compliance parameter; A correction module, used for dynamically correcting the basic peripheral vascular resistance prediction parameter and the basic arterial compliance prediction parameter according to special indicators to generate a target peripheral vascular resistance parameter and a target arterial compliance parameter; The output module is used to input the target peripheral vascular resistance parameter and the target arterial compliance parameter into the two-element Windkessel model, and calculate and generate a blood pressure estimation result based on conventional indicators, the target peripheral vascular resistance parameter and the target arterial compliance parameter.

8. A device, characterized in that The device comprises: one or more processors; a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the non-invasive blood pressure estimation method according to any one of claims 1 to 6.

9. A storage medium comprising computer-executable instructions, wherein the computer-executable instructions are used to perform the non-invasive blood pressure estimation method according to any one of claims 1 to 6 when executed by a computer processor.

10. A computer program product comprising a computer program, characterized in that: When the computer program is executed by a processor, the non-invasive blood pressure estimation method according to any one of claims 1 to 6 is implemented.

Images