A multi-source physiological data fusion analysis method for an animal model of heart failure

Abstract

This invention belongs to the field of medical data processing technology, specifically relating to a method for fusion and analysis of multi-source physiological data in an animal model of heart failure. The method includes: acquiring the raw digital arterial pressure signal sequence and echocardiographic morphological parameters of the subject; using a morphological component analysis algorithm to remove smoothing trend components to obtain a pure arterial pressure sequence; mapping the pure arterial pressure sequence to a high-dimensional phase space trajectory and calculating the pressure dissipation entropy; fusing left ventricular structural parameters, late diastolic flow velocity, and the pressure dissipation entropy to construct an impedance mismatch coupling operator; and using the impedance mismatch coupling operator to construct a calibration function to perform reverse compensation on the raw ratio, outputting an effective hemodynamic ratio. This invention can eliminate the masking effect of physiological compensation noise, avoid false normalization of data, and restore the true hemodynamic state of the subject.

Description

Technical Field

[0001] This invention relates to the field of medical data processing technology. More specifically, this invention relates to a method for the fusion and analysis of multi-source physiological data from an animal model of heart failure. Background Technology

[0002] In the field of biomedical experimental data acquisition and analysis, data processing for specific heart failure physiological models faces severe challenges. Existing measurement techniques usually rely on single-modality detection equipment and generally read sensor values ​​directly. For example, in cardiovascular assessment, parameters such as the mitral valve blood flow velocity ratio are directly measured using an ultrasound probe. Existing data processing methods mostly use conventional means such as waveform filtering to eliminate electromagnetic interference, and their underlying logic is highly dependent on ideal and constant physical conduction assumptions.

[0003] However, the above-mentioned technologies have revealed significant limitations in complex physiological environments. The raw measurement data are often inevitably mixed with nonlinear physiological compensation noise. Specifically, when an organism is in a high afterload state, its own regulatory mechanism will change the fluid dynamic environment and physical boundary conditions, resulting in a false normalization phenomenon in the ratio data collected by the sensor. That is, the measured value appears to fall into the normal range, but in reality it has been distorted and cannot objectively reflect the true blood perfusion state.

[0004] This nonlinear mismatch between data and physical state constitutes a serious signal distortion. Existing data processing methods completely lack a mechanism to quantify and remove this physiological compensatory noise. The root of the problem lies in ignoring the background constraint effect of organ structural thickening or decreased compliance on fluid fluctuations. This makes it impossible for existing analysis methods to penetrate the masking effect of the compensatory period, directly resulting in serious confidence bias in the final output physiological parameters, which can easily lead to serious misinterpretation of hemodynamic state data. Summary of the Invention

[0005] To address the technical problem of existing technologies failing to quantify and remove physiological compensatory noise, leading to spurious normalization of measurement data, this invention provides a method for fusing and analyzing multi-source physiological data from an animal model of heart failure, comprising:

[0006] The raw digital arterial pressure signal sequence and echocardiographic morphological parameters of the subject were acquired. The echocardiographic morphological parameters included left ventricular structural parameters, early diastolic flow velocity, and late diastolic flow velocity. For the raw digital arterial pressure signal sequence, a morphological component analysis algorithm based on trend and texture dictionaries was used to remove the smoothing trend component, and the retained pulsation texture component was used as the reconstructed pure arterial pressure sequence. The pure arterial pressure sequence was mapped to a high-dimensional phase space trajectory. The microstate probability density of the high-dimensional phase space trajectory in each microstate grid was calculated based on the grid partitioning method, and the pressure dissipation entropy, which characterizes the complexity of the fluid fluctuation boundary constraint, was calculated based on all the microstate probability densities. The left ventricular structural parameters, late diastolic flow velocity, and pressure dissipation entropy were fused to construct an impedance mismatch coupling operator to quantify the physiological compensation noise intensity generated by the physiological compensation mechanism of the subject. The original ratio was calculated based on the early diastolic flow velocity and the late diastolic flow velocity. The impedance mismatch coupling operator was used to construct a calibration function to reverse compensate the original ratio, and the calibrated effective hemodynamic ratio was output.

[0007] This invention acquires the original digital arterial pressure signal sequence of the subject and cardiac ultrasound morphological parameters including left ventricular structural parameters, early diastolic flow velocity, and late diastolic flow velocity. It then utilizes a morphological component analysis algorithm based on trend and texture dictionaries to remove smoothing trend components, obtaining a reconstructed pure arterial pressure sequence. This effectively eliminates additive noise such as non-physiological mechanical artifacts mixed in the original signal, providing a high-fidelity, pure data foundation for subsequent dynamic analysis. The invention maps the pure arterial pressure sequence to a high-dimensional phase space trajectory, statistically analyzes the microstate probability density based on a grid partitioning method, and calculates the pressure dissipation entropy characterizing the complexity of fluid wave boundary constraints. This allows the extraction of environmental parameters reflecting background constraints such as vascular wall compliance from the one-dimensional pressure sequence. Finally, this invention integrates left ventricular structural parameters, late diastolic flow velocity, and pressure dissipation. This invention constructs an impedance mismatch coupling operator and, through nonlinear fusion of cross-modal data, quantifies the physical environment changes caused by abnormal cardiac geometry and stiff fluid boundaries in organisms into specific physiological compensatory noise intensities. Based on the calculation of the original ratio between early diastolic flow velocity and late diastolic flow velocity, the invention utilizes the impedance mismatch coupling operator to construct a calibration function for adaptive reverse compensation of this original ratio, outputting a calibrated effective hemodynamic ratio. This eliminates the masking effect of physiological compensatory noise on the true fluid dynamic characteristics and overcomes the trap of false normalization of measurement data caused by changes in physical boundary conditions under complex physiological environments in existing technologies. It achieves accurate reconstruction of the true hemodynamic state of the subject from distorted original measurements, thus providing a high-confidence quantitative basis for the comprehensive physiological assessment of heart failure animal models.

[0008] Preferably, the step of using a morphological component analysis algorithm based on a trend dictionary and a texture dictionary to remove the smooth trend component includes: constructing a trend dictionary composed of spline functions to match low-frequency baseline drift features in the original digital arterial pressure signal sequence; constructing a texture dictionary composed of discrete cosine transforms to match high-frequency cardiovascular pulsation features in the original digital arterial pressure signal sequence; performing sparse decomposition on the original digital arterial pressure signal sequence based on the trend dictionary and the texture dictionary to separate the smooth trend component represented by the trend dictionary and the pulsation texture component represented by the texture dictionary, and removing the smooth trend component as mechanical artifact noise.

[0009] This invention constructs a trend dictionary composed of spline functions and a texture dictionary composed of discrete cosine transforms. Based on these two dictionaries, it performs sparse decomposition on the original digital arterial pressure signal sequence, separating the smooth trend component matching low-frequency baseline drift features from the pulsation texture component matching high-frequency cardiovascular pulsation features. By removing the smooth trend component, which acts as mechanical artifact noise, it effectively eliminates the severe interference of additive noise on subsequent dynamic analysis, thereby ensuring the purity of the input data and providing a high-fidelity data foundation for extracting hemodynamic features.

[0010] Preferably, mapping the pure arterial pressure sequence to a high-dimensional phase space trajectory includes: determining the delay time and embedding dimension corresponding to the pure arterial pressure sequence according to the embedding theorem; mapping the one-dimensional pure arterial pressure sequence to a high-dimensional phase space trajectory based on the delay time and the embedding dimension, wherein the high-dimensional phase space trajectory is formed by connecting multiple phase points in a multidimensional space.

[0011] Preferably, the pressure dissipation entropy satisfies the expression: In the formula, Represents the entropy of pressure dissipation; This represents the total number of microstate grids in the phase space; Indicates the first The probability density of a microstate grid; This represents the inherent electronic thermal noise figure of the sensor; Indicates the index number of the microstate grid; This represents a logarithmic function with the natural constant as its base.

[0012] This invention extracts entropy features reflecting boundary constraints using a nonlinear dynamics method to characterize the complexity of fluid fluctuations under boundary constraints. Based on this, the electronic thermal noise coefficient is introduced into the adjustment of the calculation logic, which can effectively reduce the interference of high-frequency electronic noise on entropy calculation and prevent equipment background noise from misleading the extraction of physiological features. This allows for the measurement of the background constraints of blood vessel wall state on fluid dynamics, providing reliable environmental parameters for subsequent data fusion.

[0013] Preferably, the left ventricular structural parameters include the thickness of the left ventricular posterior wall and the left ventricular end-diastolic diameter.

[0014] Preferably, the impedance mismatch coupling operator satisfies the expression: In the formula, This represents the impedance mismatch coupling operator; Indicates the thickness of the left ventricular posterior wall; Indicates the left ventricular end-diastolic diameter; This represents the geometric nonlinear weighting factor; Represents the entropy of pressure dissipation; Indicates late diastolic flow velocity; This represents the dimensional normalization factor, used to eliminate the dimension of late diastolic flow velocity.

[0015] This invention integrates left ventricular posterior wall thickness, left ventricular end-diastolic diameter, pressure dissipation entropy, and late diastolic velocity, and combines geometric nonlinear weighting factors and dimensional normalization factors to calculate the impedance mismatch coupling operator. It models two dimensions of physiological abnormalities—left ventricular structural thickening caused by abnormal cardiac geometry and decreased fluid boundary compliance caused by arteriosclerosis—as physiological compensatory noise. This can measure the degree of distortion of the original flow signal caused by changes in the complex physical environment, thereby transforming the invisible physiological compensatory effect into a calculable intensity coefficient, breaking the limitation of existing technologies that cannot isolate compensatory noise.

[0016] Preferably, the calibrated effective hemodynamic ratio satisfies the expression: In the formula, This represents the effective hemodynamic ratio after calibration; This represents the original ratio of early diastolic flow velocity to late diastolic flow velocity; This represents the impedance mismatch coupling operator; Represents the baseline noise threshold; max represents the maximum value function; This represents the natural exponential function.

[0017] This invention utilizes an impedance mismatch coupling operator, a reference noise threshold, and natural exponential and maximum functions to adaptively calibrate the original ratio of early diastolic flow velocity to late diastolic flow velocity. When the physical environment is in a reference state, the original signal measurement value remains unchanged. However, when there is severe physiological compensatory noise, the numerically suppressed original ratio is reverse-stretched to compensate. This eliminates the obscuring of the foreground measurement signal by complex background physical boundary conditions, restores the distorted physical fluid signal, and thus outputs a high-confidence effective hemodynamic ratio, avoiding misjudgment of the dynamic state caused by signal distortion.

[0018] Preferably, the reference noise threshold is obtained by: collecting physiological data of a set of reference control samples, calculating the impedance mismatch coupling operator corresponding to each sample, calculating the arithmetic mean and standard deviation of the impedance mismatch coupling operator of the reference control samples, and setting the reference noise threshold to be equal to the arithmetic mean plus 3 times the standard deviation.

[0019] Preferably, the acquisition of the original arterial pressure digital signal sequence and cardiac ultrasound morphological parameters of the subject to be tested includes: sending a synchronous trigger signal to the pressure telemetry device and the ultrasound imaging system through the main control unit, acquiring the original arterial pressure analog signal sequence and the cardiac ultrasound morphological parameters of the subject to be tested within the same time window; and performing analog-to-digital conversion on the original arterial pressure analog signal sequence to obtain a discrete original arterial pressure digital signal sequence.

[0020] Preferably, the step of calculating the microstate probability density of the high-dimensional phase space trajectory in each microstate grid based on the grid partitioning method includes: dividing the reconstructed high-dimensional phase space into multiple microstate grids of equal volume using the grid partitioning method, counting the number of phase points falling into each microstate grid, and taking the ratio of the number of phase points in each microstate grid to the total number of phase points as the microstate probability density of that microstate grid.

[0021] The beneficial effects of this invention are as follows: This invention acquires the original digital arterial pressure signal sequence and cardiac ultrasound morphological parameters of the test object, and uses a morphological component analysis algorithm based on trend and texture dictionaries to remove smooth trend components, effectively eliminating mechanical artifact noise introduced by involuntary mechanical motion, and obtaining a reconstructed pure arterial pressure sequence, thereby ensuring the high fidelity and purity of the input data. This invention maps the pure arterial pressure sequence to a high-dimensional phase space trajectory, calculates the microstate probability density and pressure dissipation entropy based on the grid partitioning method, and extracts environmental parameters reflecting fluid boundary constraints such as vascular wall compliance from the nonlinear dynamic dimension. This invention performs cross-modal data fusion of left ventricular structural parameters including left ventricular posterior wall thickness and left ventricular end-diastolic diameter, late diastolic flow velocity, and pressure dissipation entropy, constructs an impedance mismatch coupling operator, and quantifies the physical environment changes caused by abnormal cardiac geometry and fluid boundary stiffness as physiological compensatory noise intensity. This invention calculates the original ratio based on the early diastolic flow velocity and the late diastolic flow velocity, and uses the impedance mismatch coupling operator to construct an exponential calibration function to adaptively compensate the original ratio in reverse, outputting the calibrated effective hemodynamic ratio. This eliminates the masking effect of changes in complex physical boundary conditions on the foreground measurement signal, overcomes the false normalization trap caused by changes in the physiological environment during the compensatory period of heart failure in traditional measurement methods, and realizes the accurate restoration of the true hemodynamic state of the subject from the distorted original measurement values. Thus, it provides a high-confidence quantitative basis for the comprehensive analysis and evaluation of multi-source physiological data of heart failure animal models. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a method for fusing and analyzing multi-source physiological data in an animal model of heart failure according to the present invention;

[0023] Figure 2 This is a schematic diagram illustrating the effect of morphological component separation of the original digital arterial pressure signal sequence in this embodiment;

[0024] Figure 3 This is a schematic diagram comparing the original ratio of early diastolic flow velocity to late diastolic flow velocity with the calibrated effective hemodynamic ratio. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0027] This invention discloses a method for fusion analysis of multi-source physiological data in an animal model of heart failure, referring to... Figure 1 This includes steps S1-S4:

[0028] S1: Obtain the original digital arterial pressure signal sequence and cardiac ultrasound morphological parameters of the subject. The cardiac ultrasound morphological parameters include left ventricular structural parameters, early diastolic flow velocity, and late diastolic flow velocity. For the original digital arterial pressure signal sequence, use a morphological component analysis algorithm based on trend dictionary and texture dictionary to remove the smooth trend component, and use the retained pulsation texture component as the reconstructed pure arterial pressure sequence.

[0029] It should be noted that involuntary mechanical motion in the experimental environment introduces low-frequency baseline drift into the time-series signal. This additive noise can seriously interfere with subsequent dynamic analysis, making it impossible to accurately extract hemodynamic features. Therefore, this invention first ensures the spatiotemporal alignment of multi-source data through hardware synchronization at the signal acquisition end. Then, in the data preprocessing stage, the sparsity difference of signal morphology is used to remove non-physiological mechanical artifacts from the original waveform, providing a high-fidelity data foundation for subsequent feature extraction.

[0030] Specifically, the main control unit sends a synchronous trigger signal to the pressure telemetry device and the ultrasound imaging system, acquiring the original simulated arterial pressure signal sequence and cardiac ultrasound morphological parameters of the subject within the same time window. The original simulated arterial pressure signal sequence is then converted from analog to digital to obtain a discrete original digital arterial pressure signal sequence. The cardiac ultrasound morphological parameters include the left ventricular posterior wall thickness, left ventricular end-diastolic diameter, early diastolic flow velocity, and late diastolic flow velocity.

[0031] For the original digital arterial pressure (PAP) signal sequence, a trend dictionary composed of spline functions and a texture dictionary composed of discrete cosine transforms are constructed. The trend dictionary is used to match and extract low-frequency baseline drift features in the original PAP signal sequence, while the texture dictionary is used to match and extract high-frequency cardiovascular pulsation features. Using a morphological component analysis algorithm, the original PAP signal sequence is sparsely decomposed based on the trend and texture dictionaries, separating the smooth trend component represented by the trend dictionary and the pulsating texture component represented by the texture dictionary. The smooth trend component is removed as mechanical artifact noise, retaining only the pulsating texture component, which is then output as the reconstructed clean PAP sequence.

[0032] For example, Figure 2 This is a schematic diagram illustrating the effect of morphological component separation of the original digital arterial pressure signal sequence in this embodiment.

[0033] S2: Map the pure arterial pressure sequence to a high-dimensional phase space trajectory, and calculate the microstate probability density of the high-dimensional phase space trajectory in each microstate grid based on the grid partitioning method. Then, calculate the pressure dissipation entropy, which characterizes the complexity of the fluid fluctuation boundary constraint, based on all the microstate probability densities.

[0034] It should be noted that pressure values ​​alone cannot reflect the boundary conditions of the pipe in which the fluid is located, i.e., the compliance state of the blood vessel wall. The stiffness of the blood vessel wall constitutes the background constraint of fluid fluctuation. Ignoring this constraint will lead to the loss of key boundary parameters in fluid dynamics analysis. Therefore, this invention uses nonlinear dynamics methods to extract entropy features reflecting this boundary constraint from a one-dimensional pure arterial pressure sequence and uses them as environmental parameters for subsequent data fusion.

[0035] Specifically, the delay time and embedding dimension corresponding to the pure arterial pressure sequence are determined according to the embedding theorem. Based on the delay time and embedding dimension, the one-dimensional pure arterial pressure sequence is mapped to a high-dimensional phase space trajectory. The high-dimensional phase space trajectory is formed by connecting multiple phase points in a multi-dimensional space and is used to reflect the dynamic evolution characteristics of the arterial pulsation system.

[0036] The reconstructed high-dimensional phase space is divided into multiple microstate grids of equal volume using a grid partitioning method. The number of phase points falling into each microstate grid is counted, and the ratio of the number of phase points in each microstate grid to the total number of phase points is taken as the microstate probability density of that microstate grid. Based on the microstate probability densities of all microstate grids, the pressure dissipation entropy is calculated.

[0037] The pressure dissipation entropy satisfies the expression:

[0038]

[0039] In the formula, It represents the pressure dissipation entropy, which characterizes the complexity of fluid fluctuations under boundary constraints. The lower the pressure dissipation entropy, the stronger the boundary constraints, and the more rigid the system tends to be. This represents the total number of microstate grids in the phase space; Indicates the first The probability density of a microstate grid; Indicates the index number of the microstate grid; Represents a logarithmic function with the natural constant as its base; This represents the inherent electronic thermal noise figure of the sensor, when... When the denominator increases, Increasing the entropy value can reduce the interference of high-frequency electronic noise on the entropy calculation and prevent the equipment background noise from misleading the extraction of physiological features.

[0040] Regarding the electronic thermal noise figure In this embodiment, the critical experiment method is used for calibration: a simulated data environment containing strong electromagnetic interference is set up, a standard sinusoidal analog signal is connected, and the basic Shannon entropy value of the standard sinusoidal analog signal is calculated. The cutoff frequency of the digital filter is gradually increased, and the signal is dynamically loaded. The real-time Shannon entropy value after loading the signal is continuously calculated using the same method as calculating the basic Shannon entropy value. The real-time Shannon entropy value is compared with the basic Shannon entropy value. When the relative error between the real-time Shannon entropy value and the basic Shannon entropy value first exceeds a preset error threshold, this phenomenon is identified as a critical phenomenon of nonlinear divergence of entropy value due to strong electromagnetic interference. The preset error threshold is set to 5%. The ratio of the effective power of the standard sinusoidal analog signal to the effective power of the strong electromagnetic interference in the analog data environment at the time of the critical phenomenon is recorded. This ratio is used as the target signal-to-noise ratio. The reciprocal of the target signal-to-noise ratio is multiplied by a safety factor, and the product is used as the final electronic thermal noise figure. The safety factor is set to 1.2. It should be noted that the safety factor is introduced in this invention to ensure that, while preserving physiological chaotic characteristics, interference from non-biological noise is shielded to the greatest extent possible. In other embodiments, implementers can also set the electronic thermal noise factor according to the signal-to-noise ratio parameter manual of the acquisition device and laboratory electromagnetic environment monitoring data.

[0041] S3: By integrating left ventricular structural parameters, late diastolic flow velocity, and pressure dissipation entropy, an impedance mismatch coupling operator is constructed to quantify the physiological compensation noise intensity generated by the physiological compensation mechanism of the test object.

[0042] It should be noted that the distortion in the original ratio of early diastolic flow velocity to late diastolic flow velocity of mitral valve blood is due to left ventricular structural thickening caused by abnormal cardiac geometry and decreased fluid boundary compliance caused by arteriosclerosis. These two physiological abnormalities, as phenomena of the same dimension, jointly alter fluid conduction characteristics. The signal deviation caused by this change in the physical environment constitutes physiological compensatory noise. If it is not quantified, it will lead to data interpretation errors. Therefore, this invention constructs an impedance mismatch coupling operator by fusing cardiac ultrasound morphological parameters characterizing left ventricular structural thickening with pressure dissipation entropy characterizing decreased fluid boundary compliance, in order to quantify the intensity of physiological compensatory noise.

[0043] Specifically, the impedance mismatch coupling operator is calculated based on the thickness of the left ventricular posterior wall, the left ventricular end-diastolic inner diameter, the late diastolic flow velocity, and the pressure dissipation entropy.

[0044] The impedance mismatch coupling operator satisfies the expression:

[0045]

[0046] In the formula, This represents the impedance mismatch coupling operator and the intensity coefficient characterizing physiological compensation noise. The larger the value, the higher the degree of distortion of the original flow signal by the physical environment. Indicates the thickness of the left ventricular posterior wall; Indicates the left ventricular end-diastolic diameter; and The ratio of these values ​​constitutes the relative wall thickness, which characterizes the attenuation effect of the geometry on the signal. The larger the ratio, the thicker the structure. This represents a geometric nonlinear weighting factor used to adjust the contribution weight of the relative wall thickness to the impedance mismatch coupling operator. As the size increases, the influence of structural factors on noise intensity increases exponentially. This represents the pressure dissipation entropy. The smaller the value, the more rigid the fluid boundary, which leads to an increase in the value of the impedance mismatch coupling operator. This represents the late diastolic velocity, measured in centimeters per second. A smaller value indicates impaired filling, leading to an increase in the value of the impedance mismatch coupling operator. This represents the dimensionless normalization factor, with units of centimeters per second, used to eliminate the dimension of late diastolic velocity in the denominator, ensuring... It is a dimensionless numerical value.

[0047] It should be noted that during the calculation of the impedance mismatch coupling operator, when the late-diastolic velocity... When the flow rate is less than a preset threshold, the preset threshold is used instead of the late diastolic flow rate. The calculation of the impedance mismatch coupling operator is used to avoid calculation anomalies caused by a denominator of zero. The preset threshold is used to characterize the extremely weak blood perfusion state in the end-compensated stage of severe heart failure while maintaining the mathematical stability of the underlying algorithm. In this embodiment, it is set to 0.01 cm / s. In other embodiments, implementers can adaptively set this preset threshold according to the lower limit of the measurement accuracy of the ultrasound detection equipment used and the physiological flow velocity statistics of different species of heart failure animal models under extreme conditions.

[0048] For geometric nonlinear weighting factors Considering that the actual physical impedance of ventricular wall thickening during the compensated phase of heart failure is positively correlated with the cross-sectional area of ​​the thickened myocardium, and that the cross-sectional area, as a two-dimensional physical quantity, and the relative ventricular wall thickness, as a one-dimensional measurement, have a standard quadratic geometric mapping relationship, a geometric nonlinear weighting factor is used. Set to 2 to accurately map the impedance effect of abnormal cardiac geometry on signal conduction during compensated heart failure. In other embodiments, the implementer can set the geometric nonlinear weighting factor according to the species characteristics of the experimental subjects and the statistical regularity of cardiac geometry; regarding the dimensional normalization factor... In this embodiment, the value is set to 100 centimeters per second. In other embodiments, the implementer can set the dimensional normalization factor according to the order of magnitude range of the flow velocity measurement value.

[0049] S4: Calculate the original ratio based on the early diastolic flow velocity and the late diastolic flow velocity, construct a calibration function using the impedance mismatch coupling operator to perform reverse compensation on the original ratio, and output the calibrated effective hemodynamic ratio.

[0050] It should be noted that due to the masking effect of physiological compensatory noise, the original measurement data exhibits a false normalization characteristic that does not match the true physical state. Directly using this data can lead to misjudgment of the fluid dynamic state. Therefore, this invention utilizes an impedance mismatch coupling operator to construct an exponential calibration function to perform reverse compensation on the original measurement data, eliminate the masking effect of the background environment on the foreground signal, and restore the true physical signal.

[0051] Specifically, the original measured early diastolic flow velocity is obtained, the original ratio of early diastolic flow velocity to late diastolic flow velocity is calculated, and the calibrated effective hemodynamic ratio is calculated based on the original ratio and the impedance mismatch coupling operator.

[0052] The effective hemodynamic ratio satisfies the expression:

[0053]

[0054] In the formula, This represents the effective hemodynamic ratio after calibration, i.e., the true signal value after removing physiological compensation noise; This represents the original ratio of early diastolic flow velocity to late diastolic flow velocity; This represents the impedance mismatch coupling operator; Represents the baseline noise threshold; max represents the maximum value function; Representing the natural exponential function, this invention restores the original signal through the nonlinear amplification characteristics of the exponential function. Less than or equal to This indicates that the physical environment is in a baseline state. A value of 0 makes the exponent term equal to 1, resulting in an effective hemodynamic ratio. Keep the original ratio unchanged; when Significantly greater than When this occurs, it indicates the presence of severe physiological compensatory noise, with the exponential term increasing sharply, thus suppressing the numerically repressed original ratio. The blood pressure rises to a high level, thus revealing the true hemodynamic state.

[0055] For the reference noise threshold In this embodiment, the statistical distribution method is used for determination: physiological data of a set of benchmark control samples are collected, the impedance mismatch coupling operator of each sample is calculated, the arithmetic mean and standard deviation of the impedance mismatch coupling operator of the set of samples are calculated, and a benchmark noise threshold is set. The threshold is equal to the arithmetic mean plus three times the standard deviation to ensure coverage of the fluctuation range of most benchmark physical environments. In other embodiments, implementers can set the benchmark noise threshold according to the size of the benchmark sample library and the accuracy requirements of the confidence interval.

[0056] For example, Figure 3 This is a schematic diagram comparing the original ratio of early diastolic flow velocity to late diastolic flow velocity with the calibrated effective hemodynamic ratio. Figure 3 The horizontal axis represents the data sample sequence sorted by physical environment complexity, and the filled area between the original ratio curve and the calibrated effective hemodynamic ratio curve represents the amount of compensation for physiological compensatory noise.

[0057] The calibrated effective hemodynamic ratio is output as the final evaluation index of the heart failure animal model. The effective hemodynamic ratio realizes cross-modal data fusion of cardiac ultrasound morphological parameters and pure arterial pressure sequence at the underlying logic level. It overcomes the false normalization trap caused by changes in physical boundary conditions during the compensated period of heart failure in traditional single-modal measurement data. It truly and dynamically reflects the degree of diastolic function decline in the heart failure animal model, thus providing a high-confidence quantitative basis for the comprehensive analysis of multi-source physiological data of heart failure animal model, modeling quality verification, and pharmacodynamic evaluation of heart failure targeted drugs.

Claims

1. A multi-source physiological data fusion analysis method for an animal model of heart failure, characterized in that, include: The raw digital arterial pressure signal sequence and cardiac ultrasound morphological parameters of the subject were acquired. The cardiac ultrasound morphological parameters included left ventricular structural parameters, early diastolic flow velocity, and late diastolic flow velocity. For the original digital arterial pressure signal sequence, a morphological component analysis algorithm based on trend dictionary and texture dictionary is used to remove the smooth trend component, and the retained pulsation texture component is used as the reconstructed pure arterial pressure sequence. The pure arterial pressure sequence is mapped to a high-dimensional phase space trajectory. Based on a grid partitioning method, the microstate probability density of the high-dimensional phase space trajectory in each microstate grid is statistically analyzed. Then, based on all the microstate probability densities, the pressure dissipation entropy, characterizing the complexity of the fluid fluctuation boundary constraints, is calculated, satisfying the expression: ； In the formula, Represents the entropy of pressure dissipation; This represents the total number of microstate grids in the phase space; Indicates the first The probability density of a microstate grid; This represents the inherent electronic thermal noise figure of the sensor; Indicates the index number of the microstate grid; Represents a logarithmic function with the natural constant as its base; By integrating left ventricular structural parameters, late diastolic flow velocity, and pressure dissipation entropy, an impedance mismatch coupling operator is constructed to quantify the intensity of physiological compensation noise generated by the physiological compensation mechanism in the test subject. This operator satisfies the following expression: ； In the formula, This represents the impedance mismatch coupling operator; Indicates the thickness of the left ventricular posterior wall; Indicates the left ventricular end-diastolic diameter; This represents the geometric nonlinear weighting factor; Indicates late diastolic flow velocity; This represents the dimensional normalization factor, used to eliminate the dimension of late diastolic flow velocity; Based on the original ratio calculated from the early diastolic flow velocity and the late diastolic flow velocity, a calibration function is constructed using an impedance mismatch coupling operator to perform reverse compensation on the original ratio, outputting the calibrated effective hemodynamic ratio, which satisfies the expression: ； In the formula, This represents the effective hemodynamic ratio after calibration; This represents the original ratio of early diastolic flow velocity to late diastolic flow velocity; Represents the baseline noise threshold; max represents the maximum value function; This represents the natural exponential function.

2. The method for fusion analysis of multi-source physiological data in an animal model of heart failure according to claim 1, characterized in that, The algorithm for stripping smooth trend components using morphological component analysis based on trend and texture dictionaries includes: A trend dictionary composed of spline functions is constructed to match the low-frequency baseline drift features in the original digital arterial pressure signal sequence; a texture dictionary composed of discrete cosine transform is constructed to match the high-frequency cardiovascular pulsation features in the original digital arterial pressure signal sequence; based on the trend dictionary and texture dictionary, the original digital arterial pressure signal sequence is sparsely decomposed to separate the smooth trend component represented by the trend dictionary and the pulsation texture component represented by the texture dictionary, and the smooth trend component is removed as mechanical artifact noise.

3. The method for fusion analysis of multi-source physiological data in an animal model of heart failure according to claim 1, characterized in that, The mapping of the pure arterial pressure sequence to a high-dimensional phase space trajectory includes: Determine the delay time and embedding dimension corresponding to the pure arterial pressure sequence based on the embedding theorem; Based on the delay time and the embedding dimension, a one-dimensional pure arterial pressure sequence is mapped to a high-dimensional phase space trajectory, which is formed by connecting multiple phase points in a multi-dimensional space.

4. The multi-source physiological data fusion analysis method for an animal model of heart failure according to claim 1, characterized in that, The left ventricular structural parameters include the thickness of the left ventricular posterior wall and the left ventricular end-diastolic diameter.

5. The method for fusion analysis of multi-source physiological data in an animal model of heart failure according to claim 1, characterized in that, The reference noise threshold is obtained as follows: Physiological data of a set of baseline control samples are collected, the impedance mismatch coupling operator corresponding to each sample is calculated, the arithmetic mean and standard deviation of the impedance mismatch coupling operator of the baseline control samples are calculated, and the baseline noise threshold is set to be equal to the arithmetic mean plus 3 times the standard deviation.

6. The method for fusion analysis of multi-source physiological data in an animal model of heart failure according to claim 1, characterized in that, The acquisition of the raw arterial pressure digital signal sequence and cardiac ultrasound morphological parameters of the subject includes: The main control unit sends a synchronous trigger signal to the pressure telemetry device and the ultrasound imaging system, and acquires the original arterial pressure analog signal sequence and the cardiac ultrasound morphological parameters of the subject within the same time window; the original arterial pressure analog signal sequence is converted from analog to digital to obtain a discrete original arterial pressure digital signal sequence.

7. The multi-source physiological data fusion analysis method for an animal model of heart failure according to claim 1, characterized in that, The method of statistically analyzing the micro-state probability density of the high-dimensional phase space trajectory in each micro-state grid based on grid partitioning includes: The reconstructed high-dimensional phase space is divided into multiple microstate grids of equal volume using the grid partitioning method. The number of phase points falling into each microstate grid is counted, and the ratio of the number of phase points in each microstate grid to the total number of phase points is used as the microstate probability density of that microstate grid.

Images