Multi-modal data fusion analysis method and system for pregnant and birth health

Abstract

The invention provides a multimodal data fusion analysis method and system for pregnant and birth health. The method comprises the following steps: acquiring enhanced Raman spectrum data of amniotic fluid of a pregnant woman and electrical impedance spectrum data of blood of the pregnant woman; extracting a half-peak width parameter and a displacement parameter of a characteristic peak in a preset range from the enhanced Raman spectrum data, and analyzing a dielectric relaxation frequency variation parameter of a preset frequency band from the electrical impedance spectrum data; performing correlation analysis on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter; performing multi-modal data fusion analysis on the multi-dimensional interaction parameters to obtain a multi-modal data fusion analysis result including an amniotic fluid embolism probability value so as to realize risk early warning of amniotic fluid embolism; according to the multi-modal dynamic coupling and cross-scale pathological modeling, the technical barriers of data dimension limitation, static analysis lagging and multi-modal data isolation in the prior art are overcome, and an innovative technical path is provided for early accurate early warning of amniotic fluid embolism.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a multimodal data fusion analysis method and system for maternal and child health. Background Art

[0002] In the field of maternal and infant health monitoring, amniotic fluid embolism, as an obstetric emergency, is characterized by high suddenness and complex pathological mechanisms. Its early identification faces major challenges. The pathological process involves systemic inflammatory response and coagulation dysfunction caused by the entry of fetal components into the blood. Clinically, there is an urgent need for a technical solution that can integrate multimodal data in real time during prenatal examinations and intrapartum monitoring, so as to break through the limitations of single indicator detection in analyzing complex pathological mechanisms and realize dynamic assessment and stratified early warning of embolism risks.

[0003] The current mainstream technical solution adopts a monitoring mode that combines ultrasound Doppler blood flow parameters with serum inflammatory factor concentration detection. This technical solution uses ultrasound to monitor the uterine artery blood flow resistance index and placental bed blood perfusion, simultaneously detects the concentration of inflammatory factors in maternal serum, and uses a linear regression model to calculate the embolism risk assessment results. The existing technical solutions have some core defects, including relying solely on hemodynamics and serological indicators, resulting in insufficient analysis of the source of embolic substances; using a linear combination model of single measurement values, it is impossible to capture the dynamic correlation between different data, resulting in insufficient accuracy of the analysis results; there is a gap between ultrasound and serum test data in terms of spatiotemporal resolution and pathological characterization, which is prone to misjudgment due to local parameter fluctuations. Summary of the invention

[0004] The present invention provides a multimodal data fusion analysis method and system for maternal and child health, which is used to solve the problems in the prior art that the analysis data is single-dimensional, the ability to analyze the source of embolic materials is insufficient; the dynamic correlation between different data cannot be captured, resulting in insufficient accuracy of the analysis results; and misjudgment is easily caused by local parameter fluctuations.

[0005] In a first aspect, the present invention provides a multimodal data fusion analysis method for maternal and childbirth health, comprising:

[0006] Acquire enhanced Raman spectroscopy data of pregnant women's amniotic fluid and electrical impedance spectroscopy data of pregnant women's blood;

[0007] Extracting half-width parameters and displacement parameters of characteristic peaks within a preset range from the enhanced Raman spectroscopy data, and analyzing dielectric relaxation frequency variation parameters of a preset frequency band from the electrical impedance spectrum data;

[0008] Correlation analysis is performed on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter;

[0009] A multimodal data fusion analysis is performed on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve a risk warning of amniotic fluid embolism.

[0010] Optionally, the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter are subjected to correlation analysis to construct a multi-dimensional interaction parameter, including:

[0011] The half-peak width parameter is subjected to fluctuation amplitude calculation within a preset time window to generate a half-peak width fluctuation coefficient, and the displacement parameter of adjacent time segments is subjected to cumulative change rate calculation to generate a displacement cumulative change coefficient;

[0012] Performing response intensity distribution calculation on the dielectric relaxation frequency-variable parameter within a preset frequency band to generate a dielectric relaxation frequency band response intensity coefficient, and generating an extreme value distribution feature according to the dielectric relaxation frequency band response intensity coefficient;

[0013] Converting the half-peak width fluctuation coefficient and the displacement cumulative variation coefficient into a quantitative value of the fetal component release rate, and converting the extreme value distribution characteristics into a quantitative value of the immune cell migration retardation;

[0014] The multidimensional interaction parameters are constructed according to the transmembrane flux change characteristics corresponding to the quantified value of the fetal component release rate and the capillary rheological characteristics corresponding to the quantified value of the immune cell migration retardation.

[0015] Optionally, converting the half-peak width fluctuation coefficient and the displacement cumulative variation coefficient into a quantified value of the fetal component release rate, and converting the extreme value distribution characteristics into a quantified value of the immune cell migration retardation, comprises:

[0016] Calculate the osmotic pressure gradient of the placental barrier according to the change gradient of the half-peak width fluctuation coefficient within the preset time window, and calculate the transmembrane migration flux of the fetal component according to the time series change slope of the cumulative change coefficient of the displacement;

[0017] generating a quantitative value of the fetal component release rate based on the nonlinear coupling relationship between the osmotic pressure gradient and the transmembrane migration flux of the fetal component;

[0018] According to the frequency point density of the low-frequency super-threshold intensity points in the extreme value distribution characteristics, the aggregation intensity coefficient of the inflammatory factors of the placenta is calculated, and according to the maximum interval distance of the high-frequency super-threshold intensity points, the circulation inflammation diffusion resistance coefficient of the pregnant woman is calculated;

[0019] Based on the inflammatory factor aggregation intensity coefficient and the circulating inflammation diffusion resistance coefficient, combined with the geometric parameters of the uterine capillaries of pregnant women during a preset pregnancy period, a quantitative value of immune cell migration blockade is generated.

[0020] Optionally, generating a quantified value of the fetal component release rate based on the nonlinear coupling relationship between the osmotic pressure gradient and the transmembrane migration flux of the fetal component comprises:

[0021] Performing pressure flux conversion processing on the change amplitude of the osmotic pressure gradient within a preset time window and the remodeling reference radius of the uterine spiral artery of the pregnant woman during a preset pregnancy period to generate an instantaneous transmembrane pressure parameter;

[0022] Correcting the flux change rate of the transmembrane migration flux of the fetal components in adjacent time segments and the effective pore size parameters of the placental barrier to generate flux correction parameters, wherein the effective pore size parameters include the average pore size, pore size change rate and pore size distribution density of the placental barrier;

[0023] The quantified value of the fetal component release rate is generated according to a piecewise conversion rule between the instantaneous transmembrane pressure parameter and the flux correction parameter.

[0024] Optionally, a response intensity distribution calculation is performed on the dielectric relaxation frequency-variable parameter within a preset frequency band to generate a dielectric relaxation frequency band response intensity coefficient, and an extreme value distribution feature is generated according to the dielectric relaxation frequency band response intensity coefficient, including:

[0025] Dividing the dielectric relaxation frequency-variable parameter into a low-frequency sub-segment, a medium-frequency sub-segment and a high-frequency sub-segment within a preset frequency band, and calculating the intensity accumulation value of each sub-segment respectively to generate a low-frequency accumulation coefficient, a medium-frequency accumulation coefficient and a high-frequency accumulation coefficient;

[0026] The continuous frequency points where the low-frequency cumulative coefficient in the low-frequency sub-segment exceeds the first preset threshold are used as low-frequency over-threshold intensity points, and the discrete frequency points where the high-frequency cumulative coefficient in the high-frequency sub-segment exceeds the second preset threshold are used as high-frequency over-threshold intensity points;

[0027] According to the frequency point density of the low-frequency super-threshold intensity point, a low-frequency aggregation intensity factor is calculated, and according to the frequency point spacing of the high-frequency super-threshold intensity point, a high-frequency discrete intensity factor is calculated;

[0028] The low-frequency concentrated intensity factor, the high-frequency discrete intensity factor and the intermediate-frequency cumulative coefficient are superimposed to generate a dielectric relaxation frequency band response intensity coefficient;

[0029] The ratio of the distribution interval length of the low-frequency super-threshold intensity point to the total length of the low-frequency sub-segment and the ratio of the maximum interval distance of the high-frequency super-threshold intensity point to the total length of the high-frequency sub-segment are weightedly fused to generate an extreme value distribution intensity index;

[0030] According to the extreme value distribution intensity index, target feature points exceeding a preset intensity reference value are selected from the dielectric relaxation frequency band response intensity coefficient to generate extreme value distribution features.

[0031] Optionally, the multidimensional interaction parameter is constructed according to the transmembrane flux change characteristics corresponding to the quantified value of the fetal component release rate and the capillary rheological characteristics corresponding to the quantified value of the immune cell migration retardation, including:

[0032] Perform flux density correction on the flux change rate of the quantified value of the fetal component release rate within a preset time window and the effective exchange area of ​​the maternal surface of the placenta to generate a substance release density parameter;

[0033] Matching the quantified value of the immune cell migration blockage with the average blood flow velocity of the pregnant woman to generate a circulation flow resistance characteristic parameter;

[0034] According to the structural distribution characteristics of the placental villi, the substance release density parameters are weighted to generate the substance release spatial distribution parameters;

[0035] According to the topological structure of the uterine vascular network during pregnancy, the circulation flow resistance characteristic parameters are subjected to resistance gradient mapping to generate circulation resistance gradient parameters;

[0036] Generate a multidimensional interaction parameter group according to the coupling relationship between the substance release spatial distribution parameter and the circulatory resistance gradient parameter, wherein the multidimensional interaction parameter group includes a substance migration path and an immune response diffusion path at the placenta-maternal interface;

[0037] Based on the physiological benchmark parameters of pregnancy, the multidimensional interaction parameter group is range-calibrated to generate multidimensional interaction parameters. The physiological benchmark parameters of pregnancy include placental barrier function parameters and placental structure parameters.

[0038] Optionally, performing multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value to achieve risk warning of amniotic fluid embolism includes:

[0039] Based on a preset substance release density threshold, identifying a hot spot area from the substance release spatial distribution parameter in the multidimensional interaction parameter, and extracting a coordinate point corresponding to the maximum value of the substance release density from the hot spot area as a geometric center position;

[0040] Calculating the optimal diffusion path length from the geometric center position to the uterine venous return node according to the lowest resistance flow direction parameter in the circulation resistance gradient parameter;

[0041] calculating the systemic exposure of the fetal component based on the optimal diffusion path length and the release flux density of the hot spot area;

[0042] According to the ratio of the system exposure amount to the preset clearance capacity reference value, a multimodal data fusion analysis result including the probability value of amniotic fluid embolism is generated to achieve risk warning of amniotic fluid embolism.

[0043] In a second aspect, the present invention provides a multimodal data fusion analysis system for maternal and childbirth health, comprising:

[0044] A receiving module, used for acquiring enhanced Raman spectrum data of the amniotic fluid of the pregnant woman and electrical impedance spectrum data of the blood of the pregnant woman;

[0045] An extraction module, used to extract the half-peak width parameter and displacement parameter of the characteristic peak within a preset range from the enhanced Raman spectrum data, and parse the dielectric relaxation frequency variation parameter of the preset frequency band from the electrical impedance spectrum data;

[0046] A construction module is used to perform correlation analysis on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter;

[0047] The analysis module is used to perform multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve risk warning of amniotic fluid embolism.

[0048] In a third aspect, the present invention provides a computing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute a multimodal data fusion and analysis method for maternal and child health as described in any one of the first aspects.

[0049] In a fourth aspect, the present invention provides a computer storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement a multimodal data fusion and analysis method for maternal and child health as described in any one of the first aspects.

[0050] In the present invention, enhanced Raman spectroscopy data of amniotic fluid of pregnant women and electrical impedance spectrum data of blood of pregnant women are obtained; the half-width parameters and displacement parameters of characteristic peaks within a preset range are extracted from the enhanced Raman spectroscopy data, and the dielectric relaxation frequency-varying parameters of a preset frequency band are parsed from the electrical impedance spectrum data; the half-width parameters, the displacement parameters and the dielectric relaxation frequency-varying parameters are correlated and analyzed to construct multidimensional interactive parameters; multimodal data fusion analysis is performed on the multidimensional interactive parameters to obtain a multimodal data fusion analysis result including a probability value of amniotic fluid embolism, so as to realize risk warning of amniotic fluid embolism. The technical solution provided by the present invention solves the technical barrier that the existing single-modal detection (such as ultrasound or blood biochemical indicators only) cannot simultaneously capture placental material leakage and maternal immune response by synchronously collecting optical (Raman spectroscopy) and electrophysiological (impedance) dual-modal data, thereby improving the complementarity and comprehensiveness of data dimensions; accurately characterizes the release intensity and migration rate of fetal components by extracting dynamic parameters of characteristic peaks in the fingerprint area of ​​the Raman spectrum; combines the analysis of dielectric relaxation characteristics in the impedance frequency band to quantify the activation state of immune cells in pregnant women in real time, breaking through the inadequacy of the ability of traditional static parameters to analyze dynamic pathological processes; establishes a dynamic coupling relationship between optical and electrophysiological cross-modal parameters (such as the interaction between osmotic pressure gradient and migration flux), solves the problem of pathological mechanism fragmentation caused by isolated analysis of multi-source data in the existing technology, enhances the comprehensive analysis capability of the cascade effect of amniotic fluid embolism material release-immune response, and solves the problem of traditional methods ignoring the placenta-maternal The problem of early warning lag caused by the dynamic characteristics of biological processes at the placenta-maternal interface is solved, and the specificity and timeliness of early identification of amniotic fluid embolism are significantly improved. Among them, the kinetic characteristics of fetal component release are generated by the fluctuation amplitude and cumulative change rate of displacement based on the half-peak width parameters of the Raman spectrum, and the maternal inflammation diffusion characteristics are extracted by combining the response intensity distribution of the impedance dielectric relaxation frequency band; the risk of substance release is quantified by the nonlinear coupling model of the placental osmotic pressure gradient and the transmembrane migration flux, and the migration blockade of immune cells is corrected based on the capillary geometric parameters; finally, the transmembrane flux change characteristics and capillary rheological characteristics are integrated to construct multidimensional interactive parameters, breaking through the technical limitation of the existing single-modal detection technology that cannot analyze the dynamic interaction process of the placenta-maternal interface, and through the cross-modal nonlinear coupling of optical and electrophysiological parameters, the problem of misjudgment of the pathological mechanism of amniotic fluid embolism substance release-inflammatory cascade by the static correlation model in the traditional method is solved, which significantly improves the analysis accuracy and warning reliability of the dynamic evolution process of embolism risk.

[0051] These and other aspects of the present invention will become more apparent from the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 A flowchart of a multimodal data fusion analysis method for maternal and childbirth health provided by an embodiment of the present invention;

[0054] Figure 2 A schematic diagram of the structure of a multimodal data fusion analysis system for maternal and childbirth health provided by an embodiment of the present invention;

[0055] Figure 3 A schematic diagram of the structure of a computing device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0057] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0059] Figure 1 A flowchart of a multimodal data fusion analysis method for maternal and childbirth health is provided for an embodiment of the present invention, such as Figure 1 As shown, the method includes:

[0060] In response to the clinical demand for the simultaneous capture of placental material leakage and maternal immune response in early screening of amniotic fluid embolism, the existing ultrasound-serum joint detection scheme has a high false alarm rate due to the single data dimension, static parameter association and isolation of multimodal data. The present invention innovatively adopts a multimodal synergistic mechanism of enhanced Raman spectroscopy and electrical impedance spectroscopy: the dynamics of transmembrane migration of fetal components are inverted in real time through the characteristic peak parameters of the Raman spectrum, and the activation intensity of maternal immune cells is dynamically quantified in combination with the dielectric relaxation frequency variation parameters, breaking through the pathological characterization limitations of single-modal detection; further constructing cross-scale dynamic interaction parameters, and using a multimodal data spatiotemporal fusion model to analyze the pathological chain of material release to the immune cascade, and finally generating a multimodal data fusion analysis result including a dynamically evolving amniotic fluid embolism probability value, in order to solve the problem of early warning lag caused by data fragmentation and static modeling in traditional schemes, and realize a technological leap from single indicator judgment to multi-dimensional process tracking. Based on this, the present invention provides a multimodal data fusion analysis method for maternal and child health, such as Figure 1 ,include:

[0061] Step 101: Obtain enhanced Raman spectrum data of the pregnant woman's amniotic fluid and electrical impedance spectrum data of the pregnant woman's blood;

[0062] In this step, enhanced Raman spectroscopy data refers to the vibration spectrum data of amniotic fluid molecules obtained by surface enhanced Raman scattering technology, in which nano-scale metal substrates (such as gold / silver nanoparticles) are used to enhance Raman signals; this data includes the molecular fingerprint characteristics of components such as fetal squamous epithelial cells and vernix caseosa particles detected in amniotic fluid. Electrical impedance spectrum data refers to the complex impedance spectrum of pregnant women's blood measured within a preset frequency band (such as 10kHz-1MHz), reflecting the dielectric relaxation characteristics of blood cells (neutrophils, platelets), and quantifying electrophysiological parameters such as the activation state of immune cells (such as membrane capacitance changes ±15%) and aggregation degree through changes in impedance phase and amplitude at multiple frequency points.

[0063] In an embodiment of the present invention, an amniotic fluid sample of a pregnant woman is collected by aseptic amniocentesis, the sample is optically enhanced using a surface plasmon resonance (SPR) chip, a 785nm laser is used for excitation and Raman spectral signals in the range of 600-1800cm-1 are collected; a four-electrode bioimpedance meter is used simultaneously to apply a 10kHz-1MHz swept frequency current to the maternal radial artery, the blood complex impedance and phase angle are measured, and electrical impedance spectrum data are obtained.

[0064] Step 102: extracting half-width parameters and displacement parameters of characteristic peaks within a preset range from the enhanced Raman spectrum data, and analyzing dielectric relaxation frequency variation parameters of a preset frequency band from the electrical impedance spectrum data;

[0065] In this step, the characteristic peak in the preset range refers to the Raman spectrum 600-1800cm -1The characteristic vibration peaks in the fingerprint region cover the characteristic spectra of key biomarkers of amniotic fluid embolism, including 1285cm -1 (fetal squamous epithelial cell keratin CH bending vibration), 1285cm -1 (CH2 bending vibration of fetal squamous epithelial cell keratin), 1650cm -1 (Aminohydrin Amide I band). The half-peak width parameter refers to the full width at 50% of the characteristic peak intensity (FWHM), unit: cm -1 . Reflects the orderliness of the molecular environment, such as an increase in half-peak width of 5 cm -1 This indicates that the disintegration of the fetal cell membrane structure leads to spectrum broadening. The displacement parameter refers to the offset of the central wave number of the characteristic peak relative to the standard reference value (±cm -1 ), for example, squamous epithelial cell characteristic peak displacement +8cm -1 This indicates that the intracellular pH value decreases, which indicates early cell apoptosis. The dielectric relaxation frequency variation parameter refers to the change rate of the dielectric loss factor extracted in the 10kHz-100kHz (α relaxation) and 100kHz-1MHz (β relaxation) frequency bands, reflecting the changes in cell membrane polarization characteristics caused by neutrophil activation, where α relaxation corresponds to the rearrangement of membrane surface charge and β relaxation is associated with intracellular particle polarization.

[0066] In an embodiment of the present invention, Gaussian fitting is performed on the characteristic peaks of the Raman spectrum, and the half-width parameters are calculated to reflect the stability of the aggregation state of the material; the center positions of the characteristic peaks of continuous time segments are matched by a cross-correlation algorithm, and the displacement parameters are calculated to invert the migration rate of the material; the electrical impedance spectrum is fitted, and the dielectric relaxation turning point of the characteristic frequency is extracted to calculate the dielectric relaxation frequency variation parameters to reflect the diffusion rate of the inflammatory factor.

[0067] For example, extracting the 1450 cm-1 peak from the enhanced Raman spectrum of a pregnant woman at 28 weeks of pregnancy -1 The characteristic peak is at 12cm -1 Increase to 16cm -1 (volatility 33%), displacement 0.5cm -1 / min shifted to the right; impedance analysis showed that the dielectric relaxation frequency parameter dropped from 80kHz to 65kHz, indicating that the spread of inflammation was accelerated.

[0068] Step 103: performing correlation analysis on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter;

[0069] In this step, multidimensional interaction parameters refer to composite indicators that integrate optical and electrophysiological characteristics, including the osmotic pressure-flux coupling coefficient (the nonlinear product of the half-peak width fluctuation gradient and the displacement slope) and the inflammatory diffusion resistance factor (the ratio of the low-frequency dielectric loss integral to the high-frequency relaxation peak spacing, which is used to characterize the dynamic balance between the placental substance release rate and the maternal immune blockade intensity).

[0070] In an embodiment of the present invention, the half-width time series is dynamically aligned with the dielectric relaxation frequency variation parameter, and the optimal path distance is calculated as the osmotic pressure gradient indicator; the migration flux-inflammatory response coupling equation is constructed with the displacement change rate as the independent variable and the dielectric relaxation frequency variation parameter as the dependent variable; the principal component weights of the half-width fluctuation, displacement slope, and dielectric relaxation frequency variation rate are extracted through dimensionality reduction to form multidimensional interaction parameters of the imbalance of substance transport at the placenta-maternal interface and the cascade effect of systemic immune activation.

[0071] Step 104: performing multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve risk warning of amniotic fluid embolism;

[0072] In this step, the probability value of amniotic fluid embolism refers to the risk quantification output based on multidimensional interactive parameters, integrating core parameters such as placental leakage intensity, immune response delay time and hemodynamic compensation capacity.

[0073] In an embodiment of the present invention, multidimensional interactive parameters such as osmotic pressure gradient, migration flux and inflammatory diffusion resistance are input into a pre-trained dynamic coupling model, and the nonlinear interaction law between parameters is captured through the spatiotemporal feature extraction module, and probability mapping is performed in combination with the risk scale fitted by a large clinical sample; the information entropy value output by the model is calculated simultaneously to evaluate the confidence level, and when the probability value of amniotic fluid embolism exceeds the dynamic threshold and the confidence level meets the standard, a multidimensional early warning report including risk level, lesion location and re-examination suggestions is generated, thereby solving the problems of early warning delay and false alarm caused by ignoring the dynamic coupling of parameters in traditional static models.

[0074] The embodiment of the present invention breaks through the data dimension limitation of traditional ultrasonic serum detection by enhancing the dual-modal synergy of Raman spectroscopy and electrical impedance spectroscopy. Specifically, the half-width parameters and displacement parameters analyze the placental material leakage dynamics in real time, and the dielectric relaxation frequency variation parameters quantify the intensity of maternal immune response. Multi-dimensional interactive parameters are further constructed, and multi-modal spatiotemporal characteristics are integrated to generate multi-modal data fusion analysis results including dynamically evolving risk probability values. Compared with the existing technology, the problem of early warning lag caused by static correlation of single-modal parameters is solved, and the false alarm rate is reduced through the confidence assessment mechanism, providing a technical leap from single indicator judgment to multi-dimensional process tracking for amniotic fluid embolism screening.

[0075] The present invention provides a specific embodiment, step 103, performing correlation analysis on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter, specifically comprising the following steps:

[0076] Step 301: calculating the fluctuation amplitude of the half-peak width parameter within a preset time window to generate a half-peak width fluctuation coefficient, and calculating the cumulative change rate of the displacement parameter of adjacent time segments to generate a displacement cumulative change coefficient;

[0077] In this step, the fluctuation amplitude refers to the difference between the maximum and minimum values ​​of the half-peak width parameter of the Raman spectrum within the preset time window, reflecting the dynamic fluctuation intensity of the placental barrier osmotic pressure, and is used to characterize the non-steady-state characteristics of the fetal component release process. The half-peak width fluctuation coefficient refers to the quantitative index (unit: nm / min) after the fluctuation amplitude is normalized by the time window, characterizing the standardized rate of change of osmotic pressure, and is used to eliminate the interference of individual monitoring time differences on risk assessment. It is calculated by dividing the fluctuation amplitude by the length of the time window. The cumulative change rate refers to the ratio of the sum of the increments of the displacement parameters in adjacent time segments to the total time length, which is used to quantify the accelerated trend of transmembrane migration of fetal components and capture sudden events of material leakage. The cumulative change coefficient of displacement refers to the dynamic parameter of the cumulative change rate after correction by the effective exchange area of ​​the placenta, which is used to eliminate the influence of placental size differences on migration rate assessment and achieve cross-individual comparability.

[0078] In an embodiment of the present invention, the fluctuation amplitude of the half-width parameter (the difference between the maximum and minimum values ​​in the window) is extracted through a sliding time window (such as 5 minutes) to generate a half-width fluctuation coefficient; at the same time, the cumulative change rate (total displacement / total time) of the displacement parameter of adjacent time segments (such as 3 consecutive 30-second windows) is calculated to generate a cumulative displacement variation coefficient. The half-width fluctuation coefficient is used to describe the dynamic pressure fluctuation of the placental barrier, and the cumulative displacement variation coefficient is used to describe the migration rate of the substance.

[0079] Step 302: performing response intensity distribution calculation on the dielectric relaxation frequency-dependent parameters within a preset frequency band to generate a dielectric relaxation frequency band response intensity coefficient, and generating an extreme value distribution feature according to the dielectric relaxation frequency band response intensity coefficient;

[0080] In this step, the response intensity distribution refers to the signal intensity distribution of the dielectric relaxation frequency-varying parameters within a preset frequency band (such as 10kHz-1MHz), which is used to identify the characteristic frequency band response pattern of immune cell (neutrophil) activation in the blood of pregnant women. The dielectric relaxation frequency band response intensity coefficient refers to the weighted integral value of the dielectric relaxation signal intensity within a specific frequency band (such as 100-500kHz), which is used to quantify the collective effect of changes in maternal immune cell membrane capacitance and characterize the intensity of the inflammatory response. The extreme value distribution characteristics refer to the frequency point distribution characteristics (such as continuous frequency point density, maximum interval distance) that exceed the threshold in the dielectric relaxation frequency band response intensity coefficient, reflecting the spatiotemporal aggregation of immune cell activation signals, which is used to distinguish local inflammation from systemic spread.

[0081] In an embodiment of the present invention, the dielectric relaxation frequency-varying parameters of the electrical impedance spectrum are divided into low frequency (10-100kHz), medium frequency (100-500kHz), and high frequency (500kHz-1MHz), and the energy integral of each sub-frequency band is calculated to generate the response intensity distribution coefficient; the continuous high-value area of ​​the low-frequency band and the discrete peaks of the high-frequency band are screened by threshold value to generate extreme value distribution characteristics, which are used to describe the local aggregation and systemic diffusion of the maternal immune response.

[0082] Step 303: converting the half-peak width fluctuation coefficient and the displacement cumulative variation coefficient into a quantitative value of the fetal component release rate, and converting the extreme value distribution characteristics into a quantitative value of the immune cell migration retardation;

[0083] In this step, the quantified value of the fetal component release rate refers to the release rate index generated based on the nonlinear coupling of the half-peak width fluctuation coefficient (osmotic pressure gradient) and the cumulative change coefficient of displacement (migration flux), which comprehensively represents the total amount of leakage of fetal squamous epithelial cells and other components through the placental barrier per unit time. The quantified value of immune cell migration blockade refers to the flow resistance index generated based on the extreme value distribution characteristics (inflammatory factor aggregation intensity) and capillary geometric parameters (radius, length), reflecting the degree of obstruction of maternal microcirculation to neutrophil migration. The higher the value, the greater the risk of inflammatory spread.

[0084] In an embodiment of the present invention, the half-width fluctuation coefficient is input into the dynamic model of placental osmotic pressure (constructed based on the physiological parameters of pregnancy), and combined with the time slope of the cumulative change coefficient of the displacement, a quantitative value of the fetal component release rate is generated; at the same time, the low-frequency aggregation intensity and high-frequency discrete resistance in the extreme value distribution characteristics are input into the preset maternal microcirculation network model, and combined with the baseline geometric parameters of the uterine capillaries during pregnancy (such as the radius of the spiral artery), a quantitative value of the immune cell migration blockade is generated.

[0085] Step 304: constructing the multidimensional interaction parameter according to the transmembrane flux change characteristics corresponding to the quantified value of the fetal component release rate and the capillary rheological characteristics corresponding to the quantified value of the immune cell migration retardation;

[0086] In this step, the transmembrane flux variation characteristics refer to the dynamic law of the evolution of the quantitative value of the fetal component release rate over time (such as exponential growth and segmented transition), which reveals the stage characteristics of placental barrier rupture (such as early leakage and acute disintegration). Capillary rheological properties refer to the mechanical properties (such as viscosity and shear stress adaptability) exhibited by pregnant women's uterine capillaries during the migration of immune cells, which are used to quantify the physical constraints on the migration of inflammatory cells.

[0087] In an embodiment of the present invention, the transmembrane flux change characteristics are dynamically correlated with the capillary rheological characteristics through a nonlinear coupling function to generate multidimensional interaction parameters including the risk propagation path in the time and space dimensions, which are used to characterize the dynamic evolution process of the chain reaction from substance release to immune response.

[0088] The embodiment of the present invention uses Raman-impedance cross-modal parameters to capture the dynamic interaction process of placental interface material migration and pregnant woman's immune activation in real time; realizes the analysis of the complete risk evolution chain from local leakage to systemic spread; and outputs results including risk probability, spatial positioning and timely recommendations to assist doctors in making quick decisions.

[0089] The present invention provides a specific embodiment, step 303, converting the half-peak width fluctuation coefficient and the displacement cumulative variation coefficient into a quantified value of the fetal component release rate, and converting the extreme value distribution characteristics into a quantified value of the immune cell migration retardation, specifically includes the following steps:

[0090] Step 311: calculating the osmotic pressure gradient of the placental barrier according to the change gradient of the half-peak width fluctuation coefficient within a preset time window, and calculating the transmembrane migration flux of the fetal component according to the time series change slope of the cumulative change coefficient of the displacement;

[0091] In this step, the osmotic pressure gradient of the placental barrier refers to the combined pressure gradient formed by the hydrostatic pressure difference and the colloidal osmotic pressure difference between the maternal and fetal surfaces of the placenta, which is used to drive the migration of fetal components (such as squamous epithelial cells) across the placental barrier. The dynamic change characteristics of the placental barrier permeability are reflected by the inversion calculation of the time-varying gradient of the half-width fluctuation coefficient of the Raman spectrum. The temporal variation slope of the cumulative displacement coefficient refers to the rate of change of the cumulative displacement coefficient over time. It is calculated by linearly fitting the displacement data of adjacent time segments (such as a continuous 5-minute window) and is used to quantify the acceleration or slowing trend of the migration rate of fetal components. The transmembrane migration flux of fetal components refers to the total amount of fetal components (such as cell fragments, fetal fat particles) passing through the placental barrier per unit area per unit time. It is dynamically calculated by combining the osmotic pressure gradient and the migration rate, and the unit is μg / (min·cm 2 ), which directly reflects the intensity of placental material leakage.

[0092] In an embodiment of the present invention, the osmotic pressure gradient is generated by calculating the change gradient (Δ coefficient / Δt) of the half-peak width fluctuation coefficient within a continuous time window (such as 5 minutes) and combining it with the placental osmotic pressure calibration curve during pregnancy (constructed based on in vitro experimental data); at the same time, the cumulative change coefficient of the displacement is fitted with a time series slope (such as the least squares method) and combined with the effective exchange area of ​​the placenta (ultrasound measurement value) to generate the transmembrane migration flux of the fetal components, both of which respectively describe the driving force and rate of substance migration.

[0093] Step 312: generating a quantified value of the fetal component release rate based on the nonlinear coupling relationship between the osmotic pressure gradient and the transmembrane migration flux of the fetal component;

[0094] In this step, the nonlinear coupling relationship refers to the non-proportional interaction between the osmotic pressure gradient and the transmembrane migration flux of fetal components, which is manifested as an exponential increase in the migration flux when the osmotic pressure exceeds the critical threshold. It is modeled by a piecewise function or an exponential function, which is different from traditional linear regression.

[0095] In the embodiment of the present invention, a piecewise function of the osmotic pressure gradient and the transmembrane migration flux of the fetal component is established. When the osmotic pressure gradient is less than the critical threshold, the transmembrane migration flux of the fetal component is linearly related to the osmotic pressure gradient; when the osmotic pressure gradient is greater than or equal to the critical threshold, the transmembrane migration flux of the fetal component increases according to an exponential function. By dynamically calibrating the critical threshold (such as adjusting based on the gestational age), a quantitative value of the fetal component release rate per unit time and per unit placental barrier effective exchange area is generated (unit: μg / (min·cm 2 )).

[0096] Step 313: calculating the concentration intensity coefficient of inflammatory factors in the placenta according to the frequency density of the low-frequency super-threshold intensity points in the extreme value distribution characteristics, and calculating the circulation inflammation diffusion resistance coefficient of the pregnant woman according to the maximum interval distance of the high-frequency super-threshold intensity points;

[0097] In this step, the low-frequency super-threshold intensity point refers to a set of continuous frequency points in the low-frequency band (10-100kHz) electrical impedance spectrum where the dielectric relaxation response intensity exceeds the preset threshold, which characterizes the local aggregation effect of maternal immune cells (such as neutrophils) at the placental interface. The inflammatory factor aggregation intensity coefficient refers to an indicator that quantifies the spatial aggregation degree of inflammatory factors at the placental interface. It is calculated by the frequency density of the low-frequency super-threshold intensity point (the number of super-threshold points within the unit bandwidth). The higher the density, the stronger the local inflammatory response. The high-frequency super-threshold intensity point refers to a set of discrete frequency points in the high-frequency band (500kHz-1MHz) electrical impedance spectrum where the dielectric relaxation response intensity exceeds the preset threshold, reflecting the resistance characteristics of the inflammatory factor to the systemic circulation. The circulatory inflammation diffusion resistance coefficient is an indicator of the ability of the pregnant woman's systemic circulatory system to inhibit the spread of inflammation. It is calculated by the maximum interval distance of the high-frequency super-threshold intensity point (the maximum blank frequency band of adjacent super-threshold frequency points). The larger the interval, the higher the diffusion resistance.

[0098] In an embodiment of the present invention, the dielectric relaxation frequency-varying parameters of the electrical impedance spectrum are first divided into frequency bands, and a dynamic threshold (such as mean + 2 times standard deviation) is set in the low-frequency sub-segment (10-100kHz), and the number of continuous frequency points exceeding the threshold is counted and divided by the sub-segment bandwidth to generate the inflammatory factor aggregation intensity coefficient (number of points / kHz), which directly reflects the spatial aggregation degree of inflammatory factors at the placenta interface; at the same time, all super-threshold frequency points are marked in the high-frequency sub-segment (500kHz-1MHz), and the maximum interval distance between adjacent frequency points is calculated, and the ratio thereof to the total bandwidth of the high-frequency sub-segment (500kHz) is used as the circulatory inflammation diffusion resistance coefficient (dimensionless), which characterizes the diffusion obstruction of inflammatory mediators in the maternal circulation by quantifying the degree of high-frequency signal discreteness. The calculation of the two coefficients is dynamically updated using a sliding time window (such as 3 minutes) to ensure that the spatiotemporal evolution characteristics of the inflammatory response are captured in real time.

[0099] Step 314: generating a quantitative value of immune cell migration blockade based on the inflammatory factor aggregation intensity coefficient and the circulating inflammation diffusion resistance coefficient in combination with geometric parameters of uterine capillaries of pregnant women during a preset pregnancy period;

[0100] In this step, the geometric parameters refer to the anatomical characteristic parameters of the uterine spiral arteries and intervillous capillaries of pregnant women at a preset gestational period, including vascular radius (10-50 μm), length (100-500 μm) and bifurcation angle (30-60°), which are obtained through three-dimensional reconstruction of mid-pregnancy ultrasound images and used to correct the immune cell migration resistance model.

[0101] In an embodiment of the present invention, the inflammatory factor aggregation intensity coefficient and the circulating inflammation diffusion resistance coefficient are input into a preset microcirculation rheological model, and combined with the geometric parameters of uterine capillaries during pregnancy (such as the average radius of spiral arteries 12±3 μm), the critical shear stress threshold for neutrophils to pass through capillaries is calculated to generate a quantitative value (dimensionless coefficient) of immune cell migration retardation.

[0102] The embodiments of the present invention solve the core defects of the prior art (such as risk assessment based on static blood flow parameters), such as one-sided analysis of pathological mechanisms (focusing only on a single driving factor) and separation of time and space dimensions (isolated analysis of local and systemic parameters) by dynamically coupling osmotic pressure gradient with migration flux and inflammation diffusion resistance modeling; specifically, it breaks through the limitations of traditional linear regression and accurately depicts the accelerated effect of substance leakage when the placental barrier is destroyed; quantifies the migration resistance of immune cells in combination with the geometric characteristics of capillaries to improve the anatomical interpretability of inflammation diffusion risk assessment; and dynamically adjusts model parameters according to the characteristics of uterine vascular remodeling during pregnancy to avoid interference from non-pregnant physiological states.

[0103] The present invention provides a specific embodiment, step 312, based on the nonlinear coupling relationship between the osmotic pressure gradient and the transmembrane migration flux of the fetal component, generating a quantified value of the fetal component release rate, specifically comprising the following steps:

[0104] Step 321: performing pressure flux conversion processing on the change amplitude of the osmotic pressure gradient within the preset time window and the remodeling reference radius of the uterine spiral artery of the pregnant woman during the preset pregnancy period to generate an instantaneous transmembrane pressure parameter;

[0105] In this step, the remodeling reference radius refers to the standard value of the inner diameter of the uterine spiral artery during pregnancy after being remodeled by trophoblast infiltration during the process of placental formation. It reflects the degree of vascular dilation unique to pregnancy and can be used as an anatomical benchmark for pressure flux conversion and is used to standardize the differences in vascular geometry among different pregnant women. Pressure flux conversion refers to the physical process of calculating the efficiency of substance migration across the placental barrier by combining the dynamic changes in the osmotic pressure gradient with the vascular geometric parameters based on the principles of fluid mechanics. It is used to convert osmotic pressure fluctuations (dynamic input) into quantifiable pressure-driven flux parameters. The instantaneous transmembrane pressure parameter refers to the real-time calculated pressure difference index on both sides of the placental barrier, which combines the osmotic pressure gradient with the vascular resistance characteristics to characterize the driving force for transmembrane migration per unit time and per unit area, and is used to quantify the instantaneous pressure conditions for the release of fetal components.

[0106] In an embodiment of the present invention, the change amplitude of the osmotic pressure gradient in a preset time window is associated with the remodeling reference radius through a fluid mechanics model to calculate an instantaneous transmembrane pressure parameter, which dynamically reflects the actual transmembrane driving force of the placental barrier, avoiding static model deviations caused by vascular remodeling. The calculation formula of the instantaneous transmembrane pressure parameter is based on the reference osmotic pressure P0 as the base, multiplied by an exponential function of the natural constant e, the power of which is the product of the osmotic pressure gradient change amplitude ΔP and the remodeling reference radius R0, divided by the blood viscosity compensation factor; wherein the remodeling reference radius is calibrated by the Doppler ultrasound measurement data of the uterine spiral artery in the second trimester (20-24 weeks), with a value range of 12-18 μm, preferably 15 ± 2 μm.

[0107] Step 322: Correcting the flux change rate of the transmembrane migration flux of the fetal component in adjacent time segments and the effective pore size parameters of the placental barrier to generate flux correction parameters, wherein the effective pore size parameters include the average pore size, pore size change rate and pore size distribution density of the placental barrier;

[0108] In this step, the flux change rate of the transmembrane migration flux of the fetal component refers to the rate of change of the migration amount of the fetal component (such as squamous epithelial cells) through the placental barrier per unit time, reflecting the dynamic acceleration or deceleration characteristics of the substance release. The effective pore size parameter of the placental barrier refers to the functional index describing the microporous structure of the placental barrier, including the average pore size, pore size change rate and pore size distribution density of the placental barrier, which is used to correct the deviation between the theoretical flux value and the actual migration rate. The flux correction parameter refers to the dynamic calibration coefficient of the theoretical flux value through the effective pore size parameter, which reflects the time-varying characteristics of the actual permeability of the placental barrier and is used to eliminate the theoretical model error caused by the dynamic change of the barrier structure.

[0109] In an embodiment of the present invention, firstly, by comparing the difference in the transmembrane migration flux of the fetal component in adjacent time segments and dividing it by the length of the time interval, the flux change rate of the transmembrane migration flux of the fetal component in adjacent time segments is calculated. This parameter reflects the increase or decrease trend of the material migration rate per unit time, and provides basic dynamic characteristics for subsequent corrections; the flux change rate and the effective pore size parameter of the placental barrier are dynamically corrected. The specific process includes multiplying the ratio of the average pore size to the reference pore size, the resistance correction coefficient considering the pore size change rate, and the spatial weight of the pore size distribution density, and then multiplying them with the flux change rate, and finally generating a flux correction parameter that comprehensively reflects the influence of the pore size characteristics. This parameter simultaneously captures the synergistic effects of pore size, dynamic changes, and spatial distribution on the material migration efficiency.

[0110] Step 323: generating the quantified value of the fetal component release rate according to the segmented conversion rule between the instantaneous transmembrane pressure parameter and the flux correction parameter;

[0111] In this step, the segmented conversion rule refers to a strategy for calculating the fetal component release rate using different mathematical relationships (such as linear, exponential, and saturation functions) based on the threshold range of the instantaneous transmembrane pressure parameter, which is used to solve the nonlinear problem of the relationship between osmotic pressure gradient and flux.

[0112] In an embodiment of the present invention, the segmented conversion rule includes: when the instantaneous transmembrane pressure parameter exceeds a first pressure threshold, a nonlinear incremental correction of a flux correction parameter is triggered; when the instantaneous transmembrane pressure parameter is within a second pressure threshold range, a linear correction factor positively correlated with the placental villus density is applied to the flux correction parameter; according to the fetal component release rate quantification value = dynamic flux correction parameter × instantaneous transmembrane pressure parameter × time window coefficient, a fetal component release rate quantification value is generated, wherein the time window coefficient is dynamically adjusted according to the uterine contraction cycle, the uterine contraction period (pressure rise segment) coefficient is 1.2, and the interval period coefficient is 0.8; the first pressure threshold is set based on the average radius of the uterine spiral artery after remodeling during pregnancy; the second pressure threshold range is determined according to the blood flow velocity distribution on the maternal surface of the placenta.

[0113] The embodiments of the present invention solve the two major technical barriers of pressure-flux linear assumption deviation (inability to reflect the saturation effect of high-pressure area) and ignoring aperture effect (not considering the dynamic change of barrier structure) in the prior art (such as static flux calculation based on fixed osmotic pressure assumption) through dynamic pressure-flux coupling mechanism and pore multi-correction model. The core innovation lies in adaptively selecting flux conversion model according to pressure threshold to accurately match the material migration law under different states of placental barrier. The pore size, change rate and distribution density are comprehensively considered to solve the problem of migration efficiency evaluation distortion caused by single pore parameter in traditional methods. The remodeling reference radius is introduced to dynamically calibrate the pressure parameter to avoid the applicability error of the universal vascular model in pregnancy and delivery scenarios.

[0114] The present invention provides a specific embodiment, step 302, performing response intensity distribution calculation on the dielectric relaxation frequency-dependent parameter within a preset frequency band to generate a dielectric relaxation frequency band response intensity coefficient, and generating an extreme value distribution feature based on the dielectric relaxation frequency band response intensity coefficient, specifically comprising the following steps:

[0115] Step 331: Divide the dielectric relaxation frequency-variable parameter into a low-frequency sub-segment, a medium-frequency sub-segment and a high-frequency sub-segment within a preset frequency band, and calculate the intensity accumulation value of each sub-segment to generate a low-frequency accumulation coefficient, a medium-frequency accumulation coefficient and a high-frequency accumulation coefficient;

[0116] In this step, the low-frequency sub-segment refers to the division of the preset frequency band (such as 10kHz-1MHz) into a low-frequency part (such as 10kHz-100kHz), which is used to detect the polarization response characteristics of immune cells (such as neutrophils) in maternal blood. The intermediate frequency sub-segment refers to the intermediate frequency range (such as 100kHz-500kHz) in the preset frequency band, which is used to analyze the interface polarization effect of particles in the cytoplasm (such as mitochondria, lysosomes). The high-frequency sub-segment refers to the high-frequency part (such as 500kHz-1MHz) in the preset frequency band, which is used to capture the dipole relaxation characteristics of cell contents (such as DNA fragments, extracellular vesicles). The intensity accumulation value refers to the result of integrating or weighted summing the dielectric relaxation spectrum signal intensity in the sub-segment frequency domain, which characterizes the overall response intensity of the frequency band and can be calculated by frequency point intensity accumulation or area integration. The low-frequency accumulation coefficient refers to the quantitative index of the intensity accumulation value of the low-frequency sub-segment after normalization, reflecting the overall level of immune cell activation in the low-frequency range, and is used to judge the baseline intensity of the maternal immune response. The medium frequency accumulation coefficient refers to the standardized result of the intensity accumulation value of the medium frequency sub-segment, which characterizes the activity of organelles and the intensity of the release of inflammatory mediators, and is used to evaluate the cell metabolic state and local inflammatory response. The high frequency accumulation coefficient refers to the parameter after the intensity accumulation value of the high frequency sub-segment is calibrated, which reflects the intensity of cell damage or subcellular component release, and is used to detect the systemic spread risk of microparticles (such as cell fragments).

[0117] In an embodiment of the present invention, the dielectric relaxation frequency-varying parameters are divided into three frequency bands within a preset frequency band, including a low-frequency sub-segment, a medium-frequency sub-segment and a high-frequency sub-segment. For the low-frequency band (10-100kHz), the square integral of the signal amplitude is calculated to generate a low-frequency cumulative coefficient; for the medium-frequency sub-segment, the signal envelope mean is extracted to generate a medium-frequency cumulative coefficient; for the high-frequency sub-segment, the peak energy sum is statistically summed to generate a high-frequency cumulative coefficient.

[0118] Step 332: taking the continuous frequency points where the low-frequency cumulative coefficient in the low-frequency sub-segment exceeds the first preset threshold as the low-frequency over-threshold intensity points, and taking the discrete frequency points where the high-frequency cumulative coefficient in the high-frequency sub-segment exceeds the second preset threshold as the high-frequency over-threshold intensity points;

[0119] In this step, the first preset threshold refers to the critical value for judging abnormal signal strength in the low-frequency sub-segment, which is set according to the gestational age of the pregnant woman and the historical mean of the baseline low-frequency cumulative coefficient, and is used to screen the frequency points of abnormal aggregation of immune cells in the low-frequency range. The second preset threshold refers to the critical value of the intensity of identifying discrete spike signals in the high-frequency sub-segment, which is calibrated by the high-frequency noise level and pathological samples, and is used to mark isolated cell damage or microembolic events in the high-frequency range.

[0120] In an embodiment of the present invention, based on the low-frequency cumulative coefficient, an area where three or more consecutive frequency coefficients exceed the first preset threshold (T1) is identified in a low-frequency sub-segment (10-100kHz). The first preset threshold (T1) is set according to the gestational age of the pregnant woman and the historical mean of the baseline low-frequency cumulative coefficient. When the gestational age is ≥28 weeks, T1 is increased to 1.2 times the baseline value to compensate for the influence of placental maturity. For example, when the low-frequency cumulative coefficient of a certain frequency point is greater than T1 and its adjacent frequency points all meet this condition, the continuous frequency band is marked as a low-frequency super-threshold intensity point, and its starting and ending frequencies are recorded; based on the high-frequency cumulative coefficient, a single frequency coefficient is screened in a high-frequency sub-segment (500kHz-1MHz) for discrete peaks that exceed the second preset threshold (T2). For example, when the high-frequency cumulative coefficient of a certain frequency point is greater than T2 and its adjacent frequency coefficients are all lower than T2, the isolated frequency point is marked as a high-frequency super-threshold intensity point, and its frequency position is recorded.

[0121] Step 333: Calculate the low-frequency clustering intensity factor according to the frequency point density of the low-frequency super-threshold intensity point, and calculate the high-frequency discrete intensity factor according to the frequency point spacing of the high-frequency super-threshold intensity point;

[0122] In this step, the low-frequency aggregation intensity factor refers to the product of the frequency density of low-frequency super-threshold intensity points (the number of super-threshold points within a unit bandwidth) and the low-frequency cumulative coefficient, which is used to quantify the degree of local aggregation of immune cells. The high-frequency discrete intensity factor refers to the product of the inverse of the maximum interval distance of high-frequency super-threshold intensity points and the high-frequency cumulative coefficient, which characterizes the diffusion resistance of inflammatory factors.

[0123] In an embodiment of the present invention, the low-frequency aggregation intensity factor is calculated according to the calculation formula: the number of low-frequency super-threshold intensity points / total frequency width of the low-frequency sub-segment = low-frequency aggregation intensity factor; the high-frequency discrete intensity factor is calculated according to 1 / maximum frequency point spacing of high-frequency super-threshold intensity points × total frequency width of the high-frequency sub-segment = high-frequency discrete intensity factor.

[0124] Step 334: superimposing the low-frequency concentrated intensity factor, the high-frequency discrete intensity factor and the intermediate-frequency cumulative coefficient to generate a dielectric relaxation frequency band response intensity coefficient;

[0125] In an embodiment of the present invention, the calculation formula of the dielectric relaxation frequency band response intensity coefficient is: low frequency concentration intensity factor + high frequency discrete intensity factor × 0.5 + intermediate frequency cumulative coefficient = dielectric relaxation frequency band response intensity coefficient, wherein the intermediate frequency cumulative coefficient is used as a baseline adjustment item.

[0126] Step 335: weighted fusion is performed on the ratio of the distribution interval length of the low-frequency super-threshold intensity point to the total length of the low-frequency sub-segment and the ratio of the maximum interval distance of the high-frequency super-threshold intensity point to the total length of the high-frequency sub-segment to generate an extreme value distribution intensity index;

[0127] In this step, the extreme value distribution intensity index refers to the weighted fusion value of the low-frequency clustered intensity factor and the high-frequency discrete intensity factor, which comprehensively reflects the interactive effect of local immune activation and systemic inflammatory spread.

[0128] In an embodiment of the present invention, for example, the extreme value distribution intensity index = 0.7 × the distribution interval length of the low-frequency super-threshold intensity point / the total length of the low-frequency sub-segment + 0.3 × (1-the maximum interval distance of the high-frequency super-threshold intensity point / the total length of the high-frequency sub-segment), wherein the weight distribution of 0.7:0.3 is a comprehensive optimization result based on the pathological mechanism of amniotic fluid embolism (local takes precedence over systemic), clinical data verification (low-frequency prediction efficiency is higher) and technical characteristics (low-frequency stability is strong), aiming to balance the sensitivity and specificity of early warning, while reserving a technical interface for dynamic adjustment. If the pregnant woman is in the late pregnancy, the weight distribution can be adjusted to 0.6:0.4, because the risk of systemic spread increases with the enlargement of the uterus; if abnormal maternal liver enzymes are detected (indicating a decrease in clearance function), the high-frequency weight can also be temporarily increased to 0.4.

[0129] Step 336: According to the extreme value distribution intensity index, select target feature points exceeding a preset intensity reference value from the dielectric relaxation frequency band response intensity coefficient to generate an extreme value distribution feature;

[0130] In this step, the preset intensity reference value refers to the risk judgment threshold set according to the statistical distribution of the dielectric relaxation frequency band response intensity coefficient during normal pregnancy, which is used to screen abnormal feature points to distinguish physiological fluctuations from pathological signals. If the value distribution intensity index is >0.6, it is marked as a target feature point, and based on the target feature point, the extreme value distribution feature is generated.

[0131] The embodiment of the present invention solves the problem of insufficient frequency domain feature analysis in the prior art (such as single frequency band energy detection) through multi-frequency band collaborative analysis and dynamic threshold calibration.

[0132] In an embodiment of the present invention, the preset intensity reference value is the 80% quantile of the historical data. If the dielectric relaxation frequency band response intensity coefficient is greater than the preset intensity reference value and it is extremely difficult to distinguish between local aggregation and systemic diffusion) and static threshold misjudgment (ignoring differences in the development stages of inflammation), the core innovation lies in that the low-frequency, medium-frequency, and high-frequency sub-segments respectively capture the response characteristics of different immune cells to improve the accuracy of pathological pattern recognition; the spatiotemporal evolution characteristics of inflammatory signals are quantified through the extreme value distribution intensity index to avoid the limitations of a single intensity threshold in traditional methods; the extreme value distribution characteristics are generated by combining frequency domain intensity, aggregation density, and diffusion resistance to enhance the early warning capability of secondary inflammation of amniotic fluid embolism.

[0133] The present invention provides a specific embodiment, step 304, constructing the multidimensional interaction parameter according to the transmembrane flux change characteristics corresponding to the quantified value of the fetal component release rate and the capillary rheological characteristics corresponding to the quantified value of the immune cell migration retardation, specifically includes the following steps:

[0134] Step 341: performing flux density correction on the flux change rate of the quantified value of the fetal component release rate within a preset time window and the effective exchange area of ​​the maternal surface of the placenta to generate a substance release density parameter;

[0135] In this step, the flux change rate refers to the degree of dynamic change in the release rate of fetal components (such as squamous epithelial cells) across the placental barrier per unit time, reflecting the acceleration or deceleration characteristics of substance migration. The effect of placental barrier osmotic pressure fluctuations on substance migration is quantified by calculating the time series slope of the Raman spectral shift parameters. The effective exchange area of ​​the maternal surface of the placenta refers to the actual functional area of ​​contact between the placenta and the maternal blood circulation. It is calculated based on ultrasound three-dimensional reconstruction data, excluding calcification or infarction areas, and is used to standardize the substance release rate (flux per unit area) to eliminate the interference of individual placental morphological differences. Flux density correction refers to converting the original substance release rate (total flux) into a unit area flux density (such as μg / (min·cm 2 )) data processing process is used to solve the flux comparability problem caused by differences in placental size. The substance release density parameter refers to the amount of fetal component released per unit area per unit time after area correction, which characterizes the intensity of local placental leakage and is used for spatial distribution analysis to identify high-risk leakage areas.

[0136] In an embodiment of the present invention, firstly, the quantified value of the release rate of the fetal component of adjacent time segments within a preset time window is obtained, and the release rate value of the latter time segment is subtracted from the release rate value of the previous time segment to obtain the release rate change, which is then divided by the length of the time window to generate the flux change rate; then, the flux change rate is divided by the effective exchange area of ​​the maternal surface of the placenta (the actual material exchange area of ​​the chorionic membrane obtained by calculating the three-dimensional reconstruction of maternal ultrasound) to obtain a material release density parameter, which is used to characterize the migration intensity of the fetal component per unit area per unit time; for example, if the time window is 5 minutes, and the quantified value of the release rate of the fetal component increases from 10 μg / min to 15 μg / min during the period, then the flux change rate is (15-10) / 5=1 μg / min 2 , and then divided by the effective exchange area 8cm 2 The final material release density parameter is 0.125 μg / (min 2 cm 2 ).

[0137] Step 342: matching the quantified value of the immune cell migration retardation with the average blood flow velocity of the pregnant woman to generate a circulation flow resistance characteristic parameter;

[0138] In this step, the circulation flow resistance characteristic parameter refers to a composite resistance index that combines the immune cell migration blockade degree with the average maternal blood flow velocity, reflecting the ability of the maternal circulatory system to inhibit the spread of inflammation.

[0139] In an embodiment of the present invention, the average blood flow velocity of the pregnant woman is used to dynamically correct the quantified value of the immune cell migration retardation. The specific calculation process is: circulation flow resistance characteristic parameter = quantified value of the immune cell migration retardation × (1-average blood flow velocity / reference blood flow velocity), where the reference blood flow velocity is the average flow velocity of the uterine artery in mid-pregnancy, and the flow resistance effect increases when the flow velocity decreases.

[0140] Step 343: performing weight distribution on the substance release density parameter according to the structural distribution characteristics of the placental villi to generate a substance release spatial distribution parameter;

[0141] In this step, the structural distribution characteristics of the placental villi refer to the spatial arrangement density and geometric distribution pattern of the placental villi at the maternal interface, which are used for spatial weight allocation of substance release density parameters through high-frequency ultrasound or Raman spectroscopy inversion analysis, and high-density areas are given higher weights. Weight allocation refers to spatial weighting of substance release density parameters according to the villus distribution density, and the higher the density, the greater the weight. The spatial distribution parameter of substance release refers to the spatial distribution map of substance release density after the villus density weight allocation, which quantifies the leakage risk level of each area of ​​the placenta and is used to generate risk heat maps to guide clinical focus areas.

[0142] In the embodiment of the present invention, the substance release density parameter is weighted according to the structural distribution characteristics of the placental villi, wherein the weight of the high-density villi area is increased (e.g., density>120 roots / cm 2 The weight of the leakage path along the blood flow direction (ultrasound Doppler velocity measurement direction) is increased by 30%, and after the weight of the villus density and the villus arrangement direction is distributed, the material release spatial distribution parameter is generated.

[0143] Step 344: performing resistance gradient mapping on the circulation flow resistance characteristic parameters according to the topological structure of the uterine blood vessel network during pregnancy to generate circulation resistance gradient parameters;

[0144] In this step, the topological structure of the uterine vascular network during pregnancy refers to the anatomical connection relationship and spatial distribution characteristics of the uterine arteries, spiral arteries and intervillous capillaries during pregnancy, which is constructed based on the magnetic resonance imaging data or ultrasound vascular imaging data during pregnancy and used for resistance gradient mapping. The circulatory resistance gradient parameter refers to an indicator that quantifies the difference in resistance in different vascular regions, such as the ratio of primary pathway resistance, secondary pathway resistance, and tertiary pathway resistance, which is used to evaluate the diffusion efficiency of inflammatory factors from the placental interface to the systemic circulation.

[0145] In the embodiment of the present invention, according to the three-level topological structure of the uterine vascular network during pregnancy [main artery, spiral artery, and villous capillaries], combined with the hierarchical characteristics of vascular branches reconstructed from the magnetic resonance imaging data during pregnancy and childbirth, the circulation flow resistance characteristic parameters are hierarchically weighted, specifically including setting the weight of the main artery to 0.3 to reflect its physiological characteristics as a low-resistance main channel, the spiral artery is given the highest weight of 0.5 due to its key role in trophoblast infiltration and remodeling during pregnancy, and the villous capillary terminal is assigned a weight of 0.2 due to its high flow resistance characteristics. Finally, the circulation resistance gradient parameter is generated by accumulating the product of the flow resistance characteristic parameters of each level of blood vessels and their corresponding weights, that is, the circulation resistance gradient parameter = (main artery flow resistance × 0.3) + (spiral artery flow resistance × 0.5) + (villous capillary flow resistance × 0.2). This calculation quantifies the resistance contribution of different levels of blood vessels in the inflammation diffusion path and accurately characterizes the rheological characteristics of the maternal circulatory system.

[0146] Step 345: generating a multidimensional interaction parameter group according to the coupling relationship between the substance release spatial distribution parameter and the circulatory resistance gradient parameter, wherein the multidimensional interaction parameter group includes a substance migration path and an immune response diffusion path at the placenta-maternal interface;

[0147] In this step, a mathematical model describing the dynamic interaction between placental leakage intensity (substance release density) and maternal circulatory resistance distribution is developed. For example, high release density and low diffusion resistance indicate a high risk of systemic embolism; low release density and high diffusion resistance indicate that local risks are controllable. The substance migration pathway refers to the optimal diffusion trajectory of fetal components from the placental hotspot area to the entrance of the maternal circulation. It is generated based on the optimization of the resistance gradient parameters and is used to simulate the time delay effect of substance migration and guide the selection of intervention timing. The immune response diffusion pathway refers to the migration pathway of maternal immune cells (such as neutrophils) from the local aggregation area of ​​inflammation to the systemic circulation. It is generated based on the circulation resistance gradient parameters and is used to predict the diffusion direction and speed of systemic inflammatory responses (such as cytokine storms). The multidimensional interaction parameter group refers to a spatiotemporal interaction data set containing substance migration pathways and immune response diffusion pathways.

[0148] In an embodiment of the present invention, the spatial distribution parameters of substance release (reflecting the leakage intensity of the placental interface) and the circulatory resistance gradient parameters (characterizing the distribution of maternal vascular resistance) are dynamically matched through a spatial superposition algorithm, specifically including first identifying areas where the substance release density is higher than a preset threshold and the corresponding vascular resistance is lower than the critical value, marking them as high-risk substance migration paths; at the same time, screening areas with low release density but abnormally increased resistance, and marking them as immune response hysteresis areas; and constructing a multidimensional interactive parameter group based on the coupling relationship between the two, the coupling relationship of which is derived from the high leakage-low resistance pathological synergistic effect unique to pregnancy (dynamic balance between placental substance release and maternal circulatory compensation).

[0149] Step 346: Based on the physiological baseline parameters of pregnancy, the multidimensional interaction parameter group is range-calibrated to generate multidimensional interaction parameters, wherein the physiological baseline parameters of pregnancy include placental barrier function parameters and placental structure parameters;

[0150] In this step, a set of standard values ​​of physiological indicators specific to pregnancy, including placental barrier function parameters (including osmotic pressure gradient threshold, effective pore size baseline value) and placental structure parameters (including villus density distribution standard, vascular network fractal dimension), are used for dynamic calibration of multi-dimensional interactive parameters to ensure individualized adaptation of risk assessment.

[0151] In an embodiment of the present invention, the values ​​of each item in the multidimensional interactive parameter group are dynamically compared and mapped with the normal range of the physiological benchmark parameters of pregnancy: first, the intensity value of the material migration path is segmented and normalized according to the placental barrier function parameters (such as pore size distribution, osmotic pressure gradient), and then the spatial distribution of the immune response diffusion path is topologically calibrated in combination with the placental structural parameters (such as villus density, vascular fractal dimension), and finally a multidimensional interactive parameter that conforms to the physiological characteristics of pregnancy is generated. For example, when the placental pore size>15nm and the villus maturity reaches level 2, the risk level of the corresponding area is automatically increased by one level to ensure that the output results match the clinical pathological progress; when the vascular fractal dimension is <1.5, the circulatory resistance gradient parameter is corrected to 1.2 times the original value.

[0152] For example, a pregnant woman experienced decreased fetal movement at 30 weeks of pregnancy. The system performed the following analysis and found that the flux change rate increased significantly, and a high substance release density parameter was generated in combination with the effective exchange area. The average blood flow velocity of the pregnant woman was detected to have decreased, and a circulation flow resistance characteristic parameter was generated, indicating an increase in the resistance to inflammatory diffusion. The structural distribution characteristics of the placental villi detected by ultrasound showed that the density of the placental villi was uneven. After weights were assigned to the high-density areas, a substance release spatial distribution parameter biased toward the bottom of the uterus was generated. According to the topological structure of the uterine vascular network during pregnancy, the uterus showed spiral artery stenosis, and a high-resistance circulation resistance gradient parameter was generated, based on which a multidimensional interactive parameter group was constructed, which showed that there was a high-risk migration path at the bottom of the uterus, but the immune response hysteresis zone covered most of the placenta. After calibration, the multidimensional interactive parameters were generated to trigger a secondary warning, and emergency intervention was recommended.

[0153] The embodiment of the present invention solves the core defects of the existing technology (such as single-modal static model) such as insufficient spatial resolution (inability to locate leakage hot spots) and pathological-physiological disconnection (ignoring the impact of vascular remodeling during pregnancy) through spatial-rheological multi-dimensional coupling and dynamic physiological benchmark calibration. Its innovation lies in the precise correlation of substance release and inflammation diffusion path, breaking through the traditional isolated risk assessment model; based on the dynamic optimization of pregnancy-specific parameters such as fractal dimension and villus maturity, multi-dimensional interactive parameters are generated to avoid the problem of adaptability error of the general model.

[0154] The present invention provides a specific embodiment, step 104, performing multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve risk warning of amniotic fluid embolism, specifically comprising the following steps:

[0155] Step 401: Based on a preset substance release density threshold, a hot spot area is identified from the substance release spatial distribution parameters in the multi-dimensional interaction parameters, and a coordinate point corresponding to the maximum substance release density is extracted from the hot spot area as a geometric center position;

[0156] In this step, the preset substance release density threshold refers to the critical value of substance leakage per unit area of ​​the placenta interface determined by clinical cohort studies (such as μg / (min·cm 2 )) is used to distinguish normal exudation from pathological release. The threshold is adjusted dynamically with the gestational age. For example, the threshold in the third trimester is increased by 20% to compensate for changes in placental maturity. The hotspot area refers to a continuous high-density area in the spatial distribution parameters of material release that exceeds the preset threshold, reflecting abnormal local leakage at the placental-maternal interface. Its boundaries are determined by ultrasonic three-dimensional reconstruction, and the diameter is usually 2-5 cm, containing data from multiple sampling points.

[0157] In the embodiment of the present invention, firstly, in the projection space of the maternal surface of the placenta, a continuous area with a substance release density exceeding a preset substance release density threshold is screened as a hot spot area; then, the coordinates of the highest peak point of the density distribution are extracted as the geometric center position through Gaussian filtering and smoothing; for example, if the density peak of a certain area is 12 μg / (min·cm 2 ), whose coordinates (x, y) are the geometric center position. The maximum value of the material release density refers to the peak value of the release density among all sampling points in the hot spot area, which represents the most serious leakage intensity in the area. It is positively correlated with the shedding rate of fetal squamous epithelial cells through the inversion calculation of the Raman spectrum displacement parameters. The geometric center position refers to the spatial center of gravity coordinates of the hot spot area. The calculation method is to weight the coordinates of all super-threshold sampling points in the area according to the release density and mark them as the origin reference point of the material migration, for example, by positioning through the ultrasound image coordinate system (x, y, z).

[0158] Step 402: Calculate the optimal diffusion path length from the geometric center position to the uterine venous return node according to the lowest resistance flow direction parameter in the circulation resistance gradient parameter;

[0159] In this step, the lowest resistance flow direction parameter refers to the direction of the vascular branch with the least resistance in the circulation resistance gradient parameter. It is determined based on the uterine vascular network topology reconstructed based on the magnetic resonance imaging data of pregnancy and delivery, and the optimal flow direction is generated by comparing the resistance weights of the three levels of the main artery, spiral artery, and villous capillaries (such as 0.3:0.5:0.2). The uterine venous return node refers to the anatomical position where the uterine vein merges into the iliac vein in the maternal circulation. It is calibrated by Doppler ultrasound blood flow atlas and serves as the key outlet for the diffusion of substances from the placenta to the whole body. Its coordinates are determined by the three-dimensional model of the pelvic vessels.

[0160] In an embodiment of the present invention, based on the uterine vascular network topology model, the shortest path from the geometric center position to the uterine venous return node is searched along the direction of lowest resistance (such as the spiral artery trunk), and the path length is corrected according to the vascular level resistance weight (trunk weight 0.3, spiral artery 0.5, capillary 0.2), and finally the optimal diffusion path length is obtained. For example: if the trunk path accounts for 70%, the corrected length = original length × 0.3 + spiral path × 0.5 + capillary × 0.2.

[0161] Step 403: calculating the systemic exposure of the fetal component based on the optimal diffusion path length and the release flux density of the hot spot area;

[0162] In this step, the release flux density refers to the amount of material migration per unit area per unit time in the hot spot area (μg / (min·cm 2 )) is calculated by coupling the Raman spectrum half-peak width parameters with the displacement parameters and calibrated with the ultrasonic projection area.

[0163] In the embodiment of the present invention, based on the coordinates of the hot spot area in the spatial distribution parameters of the placenta-derived substance release, the unit area (1cm 2 The quantified value of the fetal component release rate (μg / min) of the fetal component was combined with the projected area ratio of the hot spot area in the maternal-placental interface (calculated by three-dimensional reconstruction of mid-pregnancy ultrasound) to generate a flux density correction factor (range 0.8-1.2); the final release flux density = release rate per unit area × flux density correction factor, unit: μg / (min·cm 2 ); the formula for calculating system exposure is: system exposure = release flux density × hot spot area × 1 / optimal diffusion path length × time window; for example, the release flux density is 10μg / (min·cm 2 ), hot spot area 5cm 2 , the optimal diffusion path length is 15cm, and the time window is 10 minutes, then the exposure amount is ≈33.3μg.

[0164] Step 404: generating a multimodal data fusion analysis result including an amniotic fluid embolism probability value according to the ratio of the system exposure amount to a preset clearance capacity reference value, so as to achieve a risk warning of amniotic fluid embolism;

[0165] In this step, the preset clearance capacity benchmark value refers to the upper limit of the ability of the pregnant woman's immune system to clear foreign substances per unit time (μg / min), which is generated through cross-modal calibration of pre-pregnancy immune cell function tests (such as neutrophil phagocytic index) and inflammatory factor levels during pregnancy (such as IL-10).

[0166] In the embodiment of the present invention, the ratio of the system exposure to the preset clearance capacity reference value is mapped to a probability value through a linear regression model. The calculation formula is: Amniotic fluid embolism probability value = 1 / (1+e -k (system exposure / preset clearance capacity benchmark value)), where k is the gestational age adjustment coefficient (k=1.2 in late pregnancy, k=1.0 in mid-pregnancy), and the preset clearance capacity benchmark value is obtained by cross-modal calibration of the phagocytic function parameters of maternal blood immune cells and the clearance rate of inflammatory factors at the placental interface.

[0167] The embodiment of the present invention solves the technical defects of the existing technology (such as single threshold alarm) such as fuzzy spatial positioning (inability to identify high-risk migration paths) and static assessment of clearance capacity (ignoring dynamic changes in the metabolic function of pregnant women) through progressive analysis of leakage source positioning - diffusion path optimization - dynamic calculation of exposure amount; its innovation lies in combining vascular resistance gradient and topological structure to accurately quantify the diffusion efficiency of embolic materials; dynamically calibrate risk thresholds based on the clearance capacity of pregnant women to avoid misjudgment caused by fixed thresholds; output results integrate spatial coordinates and timeliness recommendations to support rapid clinical decision-making.

[0168] Figure 2 A schematic diagram of a multimodal data fusion analysis system for maternal and childbirth health is provided in accordance with an embodiment of the present invention. Figure 2 As shown, the system includes:

[0169] The receiving module 21 is used to obtain enhanced Raman spectrum data of the amniotic fluid of the pregnant woman and electrical impedance spectrum data of the blood of the pregnant woman;

[0170] An extraction module 22 is used to extract the half-peak width parameter and displacement parameter of the characteristic peak within a preset range from the enhanced Raman spectrum data, and to parse the dielectric relaxation frequency variation parameter of a preset frequency band from the electrical impedance spectrum data;

[0171] A construction module 23 is used to perform correlation analysis on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter;

[0172] The analysis module 24 is used to perform multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve risk warning of amniotic fluid embolism.

[0173] Figure 2 The multimodal data fusion analysis system for maternal and child health can be performed Figure 1 The implementation principle and technical effect of the multimodal data fusion analysis method for maternal and child health described in the illustrated embodiment will not be repeated. The specific manner in which each module and unit performs operations in the multimodal data fusion analysis system for maternal and child health in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here.

[0174] In one possible design, Figure 2 The multimodal data fusion analysis system for maternal and childbirth health of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0175] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0176] The processing component 32 is used to: obtain enhanced Raman spectral data of the amniotic fluid of the pregnant woman and electrical impedance spectral data of the pregnant woman's blood; extract the half-peak width parameters and displacement parameters of the characteristic peak within a preset range from the enhanced Raman spectral data, and parse the dielectric relaxation frequency-varying parameters of the preset frequency band from the electrical impedance spectral data; perform correlation analysis on the half-peak width parameters, the displacement parameters and the dielectric relaxation frequency-varying parameters to construct multidimensional interactive parameters; perform multimodal data fusion analysis on the multidimensional interactive parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value to achieve risk warning of amniotic fluid embolism.

[0177] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.

[0178] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0179] Of course, the computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0180] The input / output interface provides an interface between the processing component and the peripheral interface module, which may be an output device, an input device, etc.

[0181] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0182] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0183] The embodiment of the present invention further provides a computer storage medium storing a computer program, which can achieve the above-mentioned Figure 1A multimodal data fusion analysis method for maternal and childbirth health in the illustrated embodiment.

[0184] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0185] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0186] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0187] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multimodal data fusion analysis method for maternal and childbirth health, characterized in that: include: Acquire enhanced Raman spectroscopy data of pregnant women's amniotic fluid and electrical impedance spectroscopy data of pregnant women's blood; Extracting half-width parameters and displacement parameters of characteristic peaks within a preset range from the enhanced Raman spectroscopy data, and analyzing dielectric relaxation frequency variation parameters of a preset frequency band from the electrical impedance spectrum data; Correlation analysis is performed on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter; A multimodal data fusion analysis is performed on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve a risk warning of amniotic fluid embolism.

2. The method according to claim 1, characterized in that The half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter are subjected to correlation analysis to construct a multi-dimensional interaction parameter, including: The half-peak width parameter is subjected to fluctuation amplitude calculation within a preset time window to generate a half-peak width fluctuation coefficient, and the displacement parameter of adjacent time segments is subjected to cumulative change rate calculation to generate a displacement cumulative change coefficient; Performing response intensity distribution calculation on the dielectric relaxation frequency-variable parameter within a preset frequency band to generate a dielectric relaxation frequency band response intensity coefficient, and generating an extreme value distribution feature according to the dielectric relaxation frequency band response intensity coefficient; Converting the half-peak width fluctuation coefficient and the displacement cumulative variation coefficient into a quantitative value of the fetal component release rate, and converting the extreme value distribution characteristics into a quantitative value of the immune cell migration retardation; The multidimensional interaction parameters are constructed according to the transmembrane flux change characteristics corresponding to the quantified value of the fetal component release rate and the capillary rheological characteristics corresponding to the quantified value of the immune cell migration retardation.

3. The method according to claim 2, characterized in that Converting the half-peak width fluctuation coefficient and the displacement cumulative variation coefficient into a quantitative value of the fetal component release rate, and converting the extreme value distribution characteristics into a quantitative value of the immune cell migration retardation, including: Calculate the osmotic pressure gradient of the placental barrier according to the change gradient of the half-peak width fluctuation coefficient within the preset time window, and calculate the transmembrane migration flux of the fetal component according to the time series change slope of the cumulative change coefficient of the displacement; generating a quantitative value of the fetal component release rate based on the nonlinear coupling relationship between the osmotic pressure gradient and the transmembrane migration flux of the fetal component; According to the frequency point density of the low-frequency super-threshold intensity points in the extreme value distribution characteristics, the aggregation intensity coefficient of the inflammatory factors of the placenta is calculated, and according to the maximum interval distance of the high-frequency super-threshold intensity points, the circulation inflammation diffusion resistance coefficient of the pregnant woman is calculated; Based on the inflammatory factor aggregation intensity coefficient and the circulating inflammation diffusion resistance coefficient, combined with the geometric parameters of the uterine capillaries of pregnant women during a preset pregnancy period, a quantitative value of immune cell migration blockade is generated.

4. The method according to claim 3, characterized in that: Based on the nonlinear coupling relationship between the osmotic pressure gradient and the transmembrane migration flux of the fetal component, a quantitative value of the fetal component release rate is generated, including: Performing pressure flux conversion processing on the change amplitude of the osmotic pressure gradient within a preset time window and the remodeling reference radius of the uterine spiral artery of the pregnant woman during a preset pregnancy period to generate an instantaneous transmembrane pressure parameter; Correcting the flux change rate of the transmembrane migration flux of the fetal components in adjacent time segments and the effective pore size parameters of the placental barrier to generate flux correction parameters, wherein the effective pore size parameters include the average pore size, pore size change rate and pore size distribution density of the placental barrier; The quantified value of the fetal component release rate is generated according to a piecewise conversion rule between the instantaneous transmembrane pressure parameter and the flux correction parameter.

5. The method according to claim 2, characterized in that: The response intensity distribution of the dielectric relaxation frequency-variable parameter is calculated within a preset frequency band to generate a dielectric relaxation frequency band response intensity coefficient, and an extreme value distribution feature is generated according to the dielectric relaxation frequency band response intensity coefficient, including: Dividing the dielectric relaxation frequency-variable parameter into a low-frequency sub-segment, a medium-frequency sub-segment and a high-frequency sub-segment within a preset frequency band, and calculating the intensity accumulation value of each sub-segment respectively to generate a low-frequency accumulation coefficient, a medium-frequency accumulation coefficient and a high-frequency accumulation coefficient; The continuous frequency points where the low-frequency cumulative coefficient in the low-frequency sub-segment exceeds the first preset threshold are used as low-frequency over-threshold intensity points, and the discrete frequency points where the high-frequency cumulative coefficient in the high-frequency sub-segment exceeds the second preset threshold are used as high-frequency over-threshold intensity points; According to the frequency point density of the low-frequency super-threshold intensity point, a low-frequency aggregation intensity factor is calculated, and according to the frequency point spacing of the high-frequency super-threshold intensity point, a high-frequency discrete intensity factor is calculated; The low-frequency concentrated intensity factor, the high-frequency discrete intensity factor and the intermediate-frequency cumulative coefficient are superimposed to generate a dielectric relaxation frequency band response intensity coefficient; The ratio of the distribution interval length of the low-frequency super-threshold intensity point to the total length of the low-frequency sub-segment and the ratio of the maximum interval distance of the high-frequency super-threshold intensity point to the total length of the high-frequency sub-segment are weightedly fused to generate an extreme value distribution intensity index; According to the extreme value distribution intensity index, target feature points exceeding a preset intensity reference value are selected from the dielectric relaxation frequency band response intensity coefficient to generate extreme value distribution features.

6. The method according to claim 2, characterized in that The multidimensional interaction parameters are constructed according to the transmembrane flux change characteristics corresponding to the quantified value of the fetal component release rate and the capillary rheological characteristics corresponding to the quantified value of the immune cell migration retardation, including: Perform flux density correction on the flux change rate of the quantified value of the fetal component release rate within a preset time window and the effective exchange area of ​​the maternal surface of the placenta to generate a substance release density parameter; Matching the quantified value of the immune cell migration blockage with the average blood flow velocity of the pregnant woman to generate a circulation flow resistance characteristic parameter; According to the structural distribution characteristics of the placental villi, the substance release density parameters are weighted to generate the substance release spatial distribution parameters; According to the topological structure of the uterine vascular network during pregnancy, the circulation flow resistance characteristic parameters are subjected to resistance gradient mapping to generate circulation resistance gradient parameters; Generate a multidimensional interaction parameter group according to the coupling relationship between the substance release spatial distribution parameter and the circulatory resistance gradient parameter, wherein the multidimensional interaction parameter group includes a substance migration path and an immune response diffusion path at the placenta-maternal interface; Based on the physiological benchmark parameters of pregnancy, the multidimensional interaction parameter group is range-calibrated to generate multidimensional interaction parameters. The physiological benchmark parameters of pregnancy include placental barrier function parameters and placental structure parameters.

7. The method according to claim 6, characterized in that Performing multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve risk warning of amniotic fluid embolism, including: Based on a preset substance release density threshold, identifying a hot spot area from the substance release spatial distribution parameter in the multidimensional interaction parameter, and extracting a coordinate point corresponding to the maximum value of the substance release density from the hot spot area as a geometric center position; Calculating the optimal diffusion path length from the geometric center position to the uterine venous return node according to the lowest resistance flow direction parameter in the circulation resistance gradient parameter; calculating the systemic exposure of the fetal component based on the optimal diffusion path length and the release flux density of the hot spot area; According to the ratio of the system exposure amount to the preset clearance capacity reference value, a multimodal data fusion analysis result including an amniotic fluid embolism probability value is generated to achieve risk warning of amniotic fluid embolism.

8. A multimodal data fusion analysis system for maternal and childbirth health, characterized in that: include: A receiving module, used for acquiring enhanced Raman spectrum data of the amniotic fluid of the pregnant woman and electrical impedance spectrum data of the blood of the pregnant woman; An extraction module, used to extract the half-peak width parameter and displacement parameter of the characteristic peak within a preset range from the enhanced Raman spectrum data, and parse the dielectric relaxation frequency variation parameter of the preset frequency band from the electrical impedance spectrum data; A construction module is used to perform correlation analysis on the half-peak width parameter, the displacement parameter and the dielectric relaxation frequency variation parameter to construct a multi-dimensional interaction parameter; The analysis module is used to perform multimodal data fusion analysis on the multidimensional interaction parameters to obtain a multimodal data fusion analysis result including an amniotic fluid embolism probability value, so as to achieve risk warning of amniotic fluid embolism.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a multimodal data fusion analysis method for maternal and childbirth health as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, a multimodal data fusion analysis method for maternal and childbirth health as described in any one of claims 1 to 7 is implemented.

Images