Physical sign monitoring data processing system after femoral artery paracentesis

By constructing a cross-entropy coupling model and a cyclic correlation entropy algorithm to analyze postoperative femoral artery puncture vital signs data, the problem of insufficient analysis of the correlation between local and global vital signs was solved, enabling accurate early warning and resource optimization, and improving the efficiency and safety of postoperative femoral artery puncture monitoring.

CN120998547APending Publication Date: 2025-11-21NANTONG TUMOR HOSPITAL
View PDF 17 Cites 0 Cited by

Patent Information

Application Number
CN202511511801.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-10-22
Publication Date
2025-11-21

AI Technical Summary

Technical Problem

Existing methods for monitoring vital signs after femoral artery puncture lack correlation analysis between local and global vital signs, cannot effectively identify the risk of complications, have inaccurate early warning mechanisms, and lack in-depth data processing, making it difficult to provide scientific intervention recommendations.

Method used

A cross-entropy coupling model is constructed to analyze the correlation between local and overall vital signs. The information transmission volume and transmission directionality index are generated through the cyclic correlation entropy algorithm. An abnormal vital sign coupling matrix is ​​established. Combined with the generalized fractal dimension calculation, the monitoring stages are divided and sensitivity analysis is performed to generate a sensitivity change feature set and realize hierarchical early warning.

Benefits of technology

It significantly improves the ability to identify early signs of postoperative complications, provides accurate early warning signals and intervention suggestions, reduces false alarm rates, optimizes the allocation of monitoring resources, and improves the efficiency of clinical monitoring.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120998547A_ABST
    Figure CN120998547A_ABST
Patent Text Reader

Abstract

The invention discloses a post-femoral artery paracentesis sign monitoring data processing system, and relates to the technical field of monitoring data processing, and the system comprises an abnormal sign analysis unit which is used for constructing a cross entropy coupling model based on the sign data of a post-femoral artery paracentesis patient so as to analyze the relevance between local signs and overall signs, obtaining an abnormal sign propagation path map; the sensitivity extraction unit is used for dividing different monitoring stages according to the abnormal sign propagation path map, performing phase feature analysis, extracting a sensitivity coefficient and generating a sensitivity change feature set; and the early warning signal generation unit is used for generating graded early warning signals and intervention suggestions based on the sensitivity change feature set according to a comparison result of the risk probability and a preset threshold value. The system has the advantages of local-overall sign correlation analysis, staged sensitivity calculation and intelligent graded early warning according to monitoring data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of monitoring data processing technology, and more specifically, to a data processing system for monitoring vital signs after femoral artery puncture. Background Technology

[0002] With the continuous rise in the incidence of cardiovascular diseases, interventional therapy has become a commonly used clinical diagnostic and treatment method. Femoral artery puncture is the most commonly used vascular access method in cardiovascular interventional therapy, with advantages such as minimal invasiveness and rapid recovery. However, postoperative complications of femoral artery puncture, such as bleeding, hematoma, and pseudoaneurysm, still occur frequently, seriously affecting patients' prognosis and quality of life. Therefore, postoperative vital sign monitoring has become a crucial link in ensuring patient safety. Traditional postoperative monitoring of femoral artery puncture mainly relies on medical staff to regularly check the puncture site and measure vital signs. By monitoring various physiological indicators of patients in real time, medical staff can promptly detect abnormalities and take corresponding intervention measures, thereby ensuring patient safety and treatment effectiveness.

[0003] Currently, the commonly used methods for monitoring vital signs after femoral artery puncture in clinical practice mainly include manual observation and electronic monitoring. Manual observation involves medical staff periodically checking the patient's puncture site, measuring vital signs, and recording relevant data. Electronic monitoring continuously records indicators such as blood pressure, heart rate, and blood oxygen saturation, and displays these data in real time on a screen. This monitoring data is usually recorded in the medical record system, and medical staff can use this data to assess the patient's recovery and determine the risk of complications. In recent years, with the development of the Internet of Things and big data technologies, some medical institutions have begun to try using intelligent monitoring systems to achieve automatic collection and preliminary analysis of patient vital sign data.

[0004] Existing methods for processing postoperative femoral artery puncture vital sign monitoring data have several shortcomings. First, current systems primarily rely on simple threshold judgments to process monitoring data, failing to effectively identify and predict potential complication risks. Second, there is a lack of correlation analysis between local vital signs (such as puncture site signs) and global vital signs (such as blood pressure and heart rate), resulting in an inability to fully extract the value of the data and a lack of comprehensive assessment of the patient's postoperative condition. Third, existing systems often overlook the characteristic differences at different monitoring stages during data processing, and the warning indicators are too rigid, easily leading to false alarms or missed alarms. Furthermore, the systems lack in-depth data mining and dynamic analysis capabilities, making it impossible to accurately calculate the sensitivity of monitoring indicators and providing healthcare professionals with tiered warnings and scientific intervention recommendations.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] To address the problems in related technologies, this invention proposes a data processing system for monitoring vital signs after femoral artery puncture. This system has the advantages of local-to-global vital sign correlation analysis, phased sensitivity calculation, and intelligent hierarchical early warning based on monitoring data, thereby solving the problems of insufficient data analysis, single monitoring indicators, and inaccurate early warning mechanisms in existing technologies.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows: A data processing system for monitoring vital signs after femoral artery puncture, comprising: The abnormal signs analysis unit is used to construct a cross-entropy coupling model based on the signs data of patients after femoral artery puncture, in order to analyze the correlation between local signs and global signs and obtain the abnormal signs propagation path map. The sensitivity extraction unit is used to divide different monitoring stages and perform phase feature analysis based on the abnormal vital signs propagation path map, extract sensitivity coefficients, and generate a sensitivity change feature set; The early warning signal generation unit is used to generate graded early warning signals and intervention suggestions based on the sensitivity change feature set and the comparison results of risk probability and preset threshold. The abnormal signs analysis unit includes: The matrix modeling module uses the cyclic correlation entropy algorithm to generate information transmission volume and transmission directionality index, and establishes anomaly characteristic coupling matrix; Among these methods, the information transmission volume and transmission directionality index are generated using the cyclic correlation entropy algorithm, and an anomaly characteristic coupling matrix is ​​established, including: Based on the local and global symptom parameter sets, the generalized fractal dimension is calculated, and the information transmission volume is generated by combining the cyclic correlation entropy algorithm. Based on the amount of information transmitted, analyze the phase synchronization and information flow direction between parameters to generate a transmission directionality index; Based on the information transmission volume and transmission directionality index, a coupling strength matrix and a polarity matrix are constructed, and an abnormal symptom coupling matrix is ​​generated through matrix operations.

[0008] Furthermore, the abnormal vital signs analysis unit is connected to the early warning signal generation unit through the sensitivity extraction unit; The abnormal signs analysis unit also includes: The parameter generation module is used to establish local and global vital sign parameter sets based on the vital sign data of patients after femoral artery puncture, and to construct a cross-entropy coupling model. The path construction module is used to build a path map of abnormal signs based on the eigenvector decomposition results of the abnormal sign coupling matrix.

[0009] Furthermore, the local vital signs parameter set includes the puncture site temperature change rate, tissue tension index, and bleeding volume measurement value; The overall set of vital signs parameters includes blood pressure fluctuation range, heart rate variability, and peripheral vascular resistance index; The expression for the cross-entropy coupling model is: ; ; In the formula, H ( L , G ) represents the cross-entropy from the local layer to the global layer of the model; P ( l i , g j ) represents the first local symptom parameter set. i The first parameter and overall vital signs parameter set j The joint probability distribution of the parameters; P ( g j | l i () indicates that the local vital signs parameters are known. l i Overall vital signs parameters under certain conditions g j The conditional probability; C This refers to the coupling coefficient from the local layer to the global layer of the model. H max This is the theoretical maximum cross-entropy value.

[0010] Furthermore, based on the local and global characteristic parameter sets, the generalized fractal dimension is calculated, and combined with the cyclic correlation entropy algorithm, the information transfer volume is generated, including: Based on a fixed time interval, the parameter sequences in the local and global vital signs parameter sets are divided into several subsequences. The local variance of each subsequence is calculated to construct a scaling transformation function. The Hearst exponent of each parameter time series is obtained by linear fitting under logarithmic coordinates. Based on the Hearst exponent, a generalized fractal dimension and quality index are constructed. By calculating the spectral width and peak position of the multifractal spectrum, the complexity of each parameter sequence and the dominant fractal characteristics are quantified. The information entropy of each parameter sequence is calculated based on the generalized fractal dimension, and then weighted and fused using the cyclic correlation entropy algorithm to generate the information transmission amount of local vital signs parameters to overall vital signs parameters.

[0011] Furthermore, the expression for the amount of information transmitted is: ; ; ; ; In the formula, Q ( l i , g j ) represents local vital signs parameters l i With overall vital signs parameters g j The amount of information transmitted between them; w 1 represents the information entropy weighting coefficient; w 2 represents the weighting coefficient of the cyclic correlation entropy; w 3 indicates the spectral width weighting coefficient; S ( q ) represents the information entropy of vital signs parameters; D q The generalized fractal dimension representing vital signs parameters; N This indicates the number of sampling points for the monitoring data; CE ( l i , g j ) represents local vital signs parameters l i With overall vital signs parameters g j Cyclic related entropy; Δ α The spectral width represents the vital signs parameter; I ( l i ) represents local vital signs parameters l i The amount of information; I ( g j ) represents overall vital signs parameters g j The amount of information; I ( l i , g j ) indicates parameter pair ( l i , g j The amount of joint information; R ( t () indicates based on observation time t The regulating factor; Δ α The spectral width represents the vital signs parameter; α max This represents the maximum value of the singularity index of the vital signs parameter; α minThis represents the minimum value of the singularity index of the vital signs parameter.

[0012] Furthermore, based on the amount of information transmitted, the phase synchronization and information flow direction between parameters are analyzed to generate a transmission directionality index, including: The parameters in the local and global characteristic parameter sets are paired up, and the Hilbert transform is performed on each pair of parameter sequences to obtain the corresponding analytical signals. The phase difference sequence of each pair of parameters is calculated and the probability distribution is statistically analyzed. The phase synchronization index is obtained by normalization to characterize the synchronization coupling strength of the parameter pair. The bidirectional information flow for each pair of parameters is calculated based on the transfer entropy, and the dominant direction of information transmission is determined by the directional index, thus obtaining the transmission directionality index.

[0013] Furthermore, the sensitivity extraction unit includes: The phase division module is used to extract the phase feature change sequence based on the temporal characteristics of the abnormal sign propagation path map, and determine the time boundaries of the acute phase, stable phase and recovery phase through dynamic time warping and threshold segmentation. The sensitivity calculation module is used to divide the monitoring stages based on time boundaries, analyze the correlation strength between the vital signs parameters in each monitoring stage, and determine the sensitivity coefficient of the monitoring indicators in each stage. The feature set generation module is used to perform time-frequency analysis on the sensitivity coefficients of monitoring indicators at each stage using a piecewise wavelet transform algorithm, and generate a sensitivity change feature set.

[0014] Furthermore, the phase segmentation module, when extracting phase feature change sequences based on the temporal characteristics of the abnormal sign propagation path map and determining the time boundaries of the acute, stable, and recovery phases through dynamic time warping and threshold segmentation, includes: Extract time series data of local-to-global sign parameter pairs from the abnormal sign propagation path map, and calculate the phase feature change sequence between parameter pairs; Based on the phase feature change sequence, a dynamic time warping algorithm is used to align the local-to-global vital signs parameters in time and generate monitoring time-series change curves. The monitoring time-series change curves are segmented using a dual-threshold iterative method to determine the time boundary points of the acute phase, stable phase, and recovery phase.

[0015] Furthermore, when the sensitivity calculation module divides the monitoring stages based on time boundaries and analyzes the correlation strength between pairs of vital signs parameters in each monitoring stage to determine the sensitivity coefficient of the monitoring indicators in each stage, it includes: Based on the time boundaries of each monitoring stage, extract local-to-global vital signs parameter pairs from time-series datasets at different stages; The correlation strength between parameter pairs at each stage was calculated, and a synchronization index matrix was constructed to quantify the degree of correlation between vital signs parameters at different stages. Based on the correlation between parameters in the synchronization index matrix, the sensitivity coefficient of each monitoring indicator relative to the overall change in vital signs is calculated.

[0016] Furthermore, the monitoring time-series change curves are segmented using a dual-threshold iterative method to determine the time boundary points for the acute phase, stable phase, and recovery phase, including: Based on the characteristics of the monitoring time-series change curve, an initial dual threshold is calculated, and the curve is initially divided into three stages: acute phase, stable phase, and recovery phase. Calculate the mean of the three stages, update the double threshold, and repeat the iterative calculation until the double threshold converges; Based on the converged double thresholds, the time boundary points of the three stages are determined, and the corresponding monitoring frequencies are established.

[0017] The beneficial effects of this invention are as follows: (1) This invention analyzes the correlation between local signs and global signs by constructing a cross-entropy coupling model, generates information transmission volume and transmission directionality index using the cyclic correlation entropy algorithm, and establishes an abnormal sign coupling matrix by combining generalized fractal dimension calculation, thereby realizing in-depth mining of patients' signs data after femoral artery puncture; at the same time, this invention can accurately capture the dynamic correlation between local sign parameters (puncture point temperature change rate, tissue tension index and bleeding volume measurement value) and global sign parameters (blood pressure fluctuation amplitude, heart rate variability and peripheral vascular resistance index), and constructs an abnormal sign propagation path map through matrix eigenvector decomposition, providing a reliable data foundation for subsequent monitoring and early warning. This analysis method based on multi-dimensional information fusion significantly improves the ability to identify early signs of postoperative complications.

[0018] (2) This invention designs a complete sensitivity extraction mechanism. It uses a dynamic time warping algorithm to align local-to-global vital signs parameters in time, uses a double threshold iterative segmentation method to determine the time boundaries of the acute phase, stable phase and recovery phase, and uses the generalized synchronization index to calculate the correlation strength between parameter pairs in each phase. Through the phased sensitivity analysis method, this invention not only achieves accurate phased classification of monitoring data, but also quantifies the correlation degree between vital signs parameters in different phases by constructing a synchronization index matrix, and then calculates the sensitivity coefficient of each monitoring indicator relative to the overall vital signs change. In addition, through time-frequency analysis using a piecewise wavelet transform algorithm, the generated sensitivity change feature set can comprehensively reflect the characteristics of vital signs changes in patients in different recovery phases, providing a more accurate quantitative basis for clinical monitoring.

[0019] (3) Based on the sensitivity change feature set, this invention establishes a scientific early warning mechanism. By training the risk probability model through the support vector machine algorithm and combining it with the preset multi-level early warning thresholds, the system realizes the automatic classification of early warning signals. At the same time, the system can issue early warning signals of different levels in a timely manner according to the calculation results of the risk probability and generate corresponding intervention suggestions. This data-driven dynamic early warning mechanism not only improves the accuracy of early warning of postoperative complications, but also realizes the rational allocation of monitoring resources by adaptively adjusting the monitoring frequency, providing scientific decision support for medical staff. In addition, the system's graded early warning function can effectively reduce the false alarm rate, avoid excessive consumption of medical resources, and improve the overall efficiency of clinical monitoring. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of a data processing system for monitoring vital signs after femoral artery puncture according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the working principle of a femoral artery puncture postoperative vital sign monitoring data processing system according to an embodiment of the present invention. Detailed Implementation

[0022] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention. The components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0023] According to an embodiment of the present invention, a data processing system for monitoring vital signs after femoral artery puncture is provided.

[0024] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 and Figure 2 As shown, according to an embodiment of the present invention, a femoral artery puncture postoperative vital sign monitoring data processing system is provided, the femoral artery puncture postoperative vital sign monitoring data processing system comprising: Abnormal sign analysis unit 1 is used to construct a cross-entropy coupling model based on the sign data of patients after femoral artery puncture, so as to analyze the correlation between local signs and global signs and obtain the abnormal sign propagation path map. Sensitivity extraction unit 2 is used to divide different monitoring stages and perform phase feature analysis based on the abnormal sign propagation path map, extract sensitivity coefficients, and generate a sensitivity change feature set; The early warning signal generation unit 3 is used to generate graded early warning signals and intervention suggestions based on the sensitivity change feature set and the comparison results of risk probability and preset threshold. In one embodiment, the abnormal physical sign analysis unit 1 is connected to the early warning signal generation unit 3 through the sensitivity extraction unit 2; Abnormal signs analysis unit 1 includes: The parameter generation module 101 (not shown in the figure) is used to establish a local sign parameter set and a global sign parameter set based on the sign data of patients after femoral artery puncture, and to construct a cross-entropy coupling model.

[0025] Specifically, vital sign data of patients after femoral artery puncture are collected, and local vital sign parameters are set. L ={ l 1, l 2, l 3} and overall vital sign parameter set G ={ g 1, g 2, g 3} is normalized, where l 1 represents the rate of temperature change at the puncture site. l 2 represents the tissue tension index. l 3 represents the measured value of oozing blood. g 1 represents the amplitude of blood pressure fluctuation. g 2 represents heart rate variability. g 3 represents the peripheral vascular resistance index.

[0026] Specifically, the local vital signs parameter set includes the puncture site temperature change rate, tissue tension index, and bleeding volume measurement; the overall vital signs parameter set includes blood pressure fluctuation amplitude, heart rate variability, and peripheral vascular resistance index.

[0027] Specifically, the expression for the cross-entropy coupling model is: ; ; In the formula, H ( L , G ) represents the cross-entropy from the local layer to the global layer of the model; P ( l i ,g j ) represents the first local symptom parameter set. i The first parameter and overall vital signs parameter set j The joint probability distribution of the parameters; P ( g j | l i () indicates that the local vital signs parameters are known. l i Overall vital signs parameters under certain conditions g j The conditional probability; C This refers to the coupling coefficient from the local layer to the global layer of the model. H max This is the theoretical maximum cross-entropy value.

[0028] Specifically, H max =-log(1 / J) is the theoretical maximum cross-entropy value, and J is the number of parameters in the overall trait parameter set (J=3 in this embodiment). When the parameters of the local layer and the overall layer are completely independent, i.e., P( g j | l i When )=1 / N, cross-entropy H ( L , G (reaching maximum value) H max The larger the C value, the higher the coupling degree between local signs and overall signs.

[0029] In one embodiment, the abnormal physical signs analysis unit 1 further includes a matrix modeling module 102 (not shown in the figure), which uses the cyclic correlation entropy algorithm to generate information transmission volume and transmission directionality index, and establishes an abnormal physical signs coupling matrix. When the matrix modeling module 102 uses the cyclic correlation entropy algorithm to generate the information transfer volume and the transfer directionality index, and establishes the anomaly characteristic coupling matrix, it includes: S121. Calculate the generalized fractal dimension based on the local and global characteristic parameter sets, and generate the information transmission volume by combining the cyclic correlation entropy algorithm. S122. Based on the amount of information transmitted, analyze the phase synchronization and information flow direction between parameters, and generate a transmission directionality index. S123. Based on the information transmission volume and transmission directionality index, construct the coupling strength matrix and polarity matrix, and generate the abnormal symptom coupling matrix through matrix operations.

[0030] In one embodiment, the generalized fractal dimension is calculated based on the local and global characteristic parameter sets, and combined with the cyclic correlation entropy algorithm, the information transfer quantity is generated, including: S1211. Based on a fixed time interval, divide the parameter sequences of the local and global vital signs parameter sets into several subsequences, calculate the local variance for each subsequence to construct a scaling transformation function, and obtain the Hearst exponent of each parameter time series through linear fitting in logarithmic coordinates. S1212. Based on the Hearst exponent, a generalized fractal dimension and quality index are constructed. By calculating the spectral width and peak position of the multifractal spectrum, the complexity of each parameter sequence and the dominant fractal characteristics are quantified. S1213. Calculate the information entropy of each parameter sequence based on the generalized fractal dimension, and perform weighted fusion by combining it with the cyclic correlation entropy algorithm to generate the information transmission amount of local vital signs parameters to overall vital signs parameters.

[0031] Specifically, fractal dimension analysis preprocessing is performed on the time series of the local characteristic parameter set L={l1,l2,l3} and the global characteristic parameter set G={g1,g2,g3} in the two-layer cross-entropy coupling model constructed by parameter generation module 101: First, each parameter series is divided into N subsequences according to a fixed time interval τ, and the local variance σ(τ) is calculated for each subsequence; then, a scaling transformation function F(τ)=σ(τ) / τ^α is constructed, where α is the scaling exponent; finally, the Hearst exponent H is obtained by linear fitting logF(τ)=H·logτ+C under logarithmic coordinates, where the correlation coefficient R is determined by calculation. 2 To assess the goodness of fit, R 2 The closer the R value is to 1, the better the fit. 2 A value ≥0.9 is considered to have good fractal properties.

[0032] Specifically, based on the obtained Hearst exponent, the multifractal spectrum method is used to deeply analyze the fractal characteristics of the parameter sequence: first, the generalized fractal dimension D is constructed. q =(τ(q)-1) / (q-1), where q is any real number and τ(q) is the quality index; then calculate the multifractal spectrum f(α)=qα-τ(q), where α is the singularity index; finally, use the spectral width Δα=α max -α min The complexity and dominant fractal characteristics of the spectral peak position α0 quantized signal; In one embodiment, the expression for the amount of information transmitted is: ; ; ; ; In the formula, Q ( l i , g j ) represents local vital signs parameters l i With overall vital signs parameters g j The amount of information transmitted between them; w 1 represents the information entropy weighting coefficient; w 2 represents the weighting coefficient of the cyclic correlation entropy; w 3 indicates the spectral width weighting coefficient; S ( q ) represents the information entropy of vital signs parameters; D q The generalized fractal dimension representing vital signs parameters; N This indicates the number of sampling points for the monitoring data; CE ( l i , g j ) represents local vital signs parameters l i With overall vital signs parameters g j Cyclic related entropy; Δ α The spectral width represents the vital signs parameter; I ( l i ) represents local vital signs parameters l i The amount of information; I ( g j ) represents overall vital signs parameters g j The amount of information; I ( l i , g j ) indicates parameter pair ( l i , g j The amount of joint information; R ( t () indicates based on observation time t The regulating factor; Δ α The spectral width represents the vital signs parameter; α max This represents the maximum value of the singularity index of the vital signs parameter; α min This represents the minimum value of the singularity index of the vital signs parameter.

[0033] Specifically, by combining fractal analysis results with cyclic correlation entropy, an evaluation model for the information transfer between parameters is constructed: firstly, based on the generalized fractal dimension... Dq Calculate the information entropy S(q) of the parameter sequence. D q ×ln( N Then, the cyclic correlation entropy CE is introduced. l i , g j The dynamic correlation of the quantified parameter pairs is then obtained through weighted fusion to obtain the information transfer quantity Q( l i , g j ).

[0034] Specifically, the determination of the weight coefficients in this invention is based on an adaptive optimization algorithm. First, the objective function J( w 1, w 2, w 3), this function characterizes the amount of information transmitted Q( l i , g j The accuracy of characterization of the correlation between local and global symptom parameters: The information entropy S(q) obtained by calculating the generalized fractal dimension reflects the intrinsic complexity of the parameter sequence, and the cyclic correlation entropy CE(q) reflects the accuracy of characterization of the correlation between local and global symptom parameters. l i , g j The dynamic coupling strength of the parameter pair is characterized by the spectral width Δ. α Multifractal properties of quantized signals. ① Taking the partial derivative of the objective function yields the gradient descent direction: ② An iterative optimization algorithm is adopted, with the initial weights all set to 1 / 3. In each iteration, the weight coefficients are adjusted according to the change in the objective function value. ③ When the difference between the objective function values ​​of two consecutive iterations is less than the preset threshold ε (ε=10), the algorithm is optimized. -6 When the iteration stops, it eventually converges. w 1 = 0.4 w 2 = 0.4 w 3 = 0.2. This data-driven weight optimization method avoids the interference of subjective factors and ensures the objectivity and accuracy of information transmission calculation.

[0035] Specifically, taking the monitoring data of a 65-year-old female patient within 2 hours after femoral artery puncture as an example, the local temperature change rate l1 was first preprocessed using fractal methods, and the data was divided into 24 subsequences at 5-minute intervals. Through scaling function analysis, a linear fitting equation of logF(τ) = 0.78logτ + 0.25 was obtained in log-log coordinates, with a correlation coefficient R0. 2=0.96, indicating that the parameter sequence has good fractal properties, and the Hearst exponent H=0.78.

[0036] Specifically, further multifractal analysis is performed, taking q∈[-5,5] to calculate the generalized fractal dimension, obtaining the multifractal spectrum f(α), and the spectral width Δ α =0.45, spectral peak position α α0 = 1.22. The large spectral width indicates that the temperature change has significant multifractal characteristics, while α0 > 1 indicates that the signal shows a continuous increasing trend.

[0037] Specifically, finally, the information entropy S(q) = 2.34 of the local temperature change rate l1 was calculated, along with the blood pressure fluctuation amplitude. g The cyclic correlation entropy CE(l1,g1) = 0.72. Weighting coefficients are used. w 1 = 0.4 w 2 = 0.4 w 3 = 0.2, and the information transmission volume Q is calculated. l 1, g 1) = 0.82. This value is significantly higher than the average transfer of 0.45 for other parameter pairs, indicating a strong correlation between local temperature changes and blood pressure fluctuations. This result was input into the abnormal sign coupling matrix M, providing an important basis for subsequent complication warning. In the subsequent monitoring of this patient, the system successfully predicted abnormal blood pressure fluctuations based on this strong correlation, enabling medical staff to promptly detect early signs of puncture site bleeding.

[0038] In one embodiment, generating a transmission directionality index by analyzing the phase synchronization and information flow direction between parameters based on the amount of information transmitted includes: S1221. Pair the parameters in the local and global characteristic parameter sets one by one, and perform Hilbert transform on each pair of parameter sequences to obtain the corresponding analytical signal. S1222. Calculate the phase difference sequence of each pair of parameters and statistically analyze the probability distribution. Obtain the phase synchronization index through normalization to characterize the synchronization coupling strength of the parameter pair. S1223. Calculate the bidirectional information flow for each pair of parameters based on the transfer entropy, determine the dominant direction of information transmission through the directional index, and obtain the transmission directionality index.

[0039] Specifically, taking the monitoring data of a 65-year-old female patient as an example, based on the information transmission matrix obtained by S121, phase reset detection analysis was performed on the parameter pair with the largest information transmission (Q(l1,g1)=0.82) of local temperature change rate l1 and blood pressure fluctuation amplitude g1.

[0040] Specifically, the two parameter sequences are first transformed using a Hilbert transform to obtain an analytic signal, expressed as: z1(t)=l1(t)+i×H[l1(t)]; z2(t)=g1(t)+i×H[g1(t)]; In the formula, H[·] represents the Hilbert transform; i is the imaginary unit (i 2 =-1); the analytic signals z1(t) and z2(t) contain the real part of the original signal and the imaginary part of its Hilbert transform. This complex signal representation makes it possible to accurately calculate the instantaneous phase and amplitude of the signal.

[0041] Specifically, the phase difference sequence Δφ(t) = φ1(t) - φ2(t) is then calculated, where φ1(t) and φ2(t) are the instantaneous phases of z1(t) and z2(t), respectively. By statistically analyzing the probability distribution P(Δφ) of the phase difference, the phase synchronization index γ = 0.72 is calculated, indicating that the two parameters have strong phase synchronization.

[0042] Specifically, the transfer entropy is finally calculated to determine the direction of information flow, and the expression is: TE(l1→g1)=0.95 (the amount of information transmitted from local temperature to blood pressure); TE(g1→l1)=0.12 (the amount of information transmitted from blood pressure to local temperature). The directional index D = TE(l1→g1) - TE(g1→l1) = 0.83; Positive values ​​indicate that information mainly flows from the rate of change in local temperature to the amplitude of blood pressure fluctuations, verifying that local temperature changes can serve as a leading indicator of abnormal blood pressure fluctuations.

[0043] In one embodiment, a coupling strength matrix and a polarity matrix are constructed based on the amount of information transmitted and the directionality index of transmission, and an abnormality coupling matrix is ​​generated through matrix multiplication.

[0044] Specifically, a two-layer network model was constructed based on the above analysis results. Three nodes were set in the local signs layer: local temperature change rate (l1), tissue tension index (l2), and bleeding volume measurement value (l3); three nodes were set in the global signs layer: blood pressure fluctuation amplitude (g1), heart rate variability (g2), and peripheral vascular resistance index (g3).

[0045] Specifically, first, an inter-layer connection weight matrix W is established, whose element values ​​are the information transmission amount of the corresponding parameter pairs: W(l1,g1)=0.82, W(l1,g2)=0.45, W(l1,g3)=0.38; W(l2,g1)=0.41, W(l2,g2)=0.35, W(l2,g3)=0.65; W(l3,g1)=0.39, W(l3,g2)=0.42, W(l3,g3)=0.44.

[0046] Specifically, the connection polarity P is then determined based on the transmission directionality index D: P(l1,g1)=+1 (D=0.83>0); P(l1,g2)=-1 (D=-0.25<0); The polarity of other parameter pairs follows the same logic.

[0047] Specifically, the resulting abnormal symptom coupling matrix M is formed: M(i,j)=W(i,j)×P(i,j); By analyzing the characteristic structure of the M matrix, we can identify: The strongest coupling path is l1→g1 (weight 0.82, positive). Second strongest coupling path: l2→g3 (weight 0.65, positive); Feedback coupling loop: l1→g1→l3; Based on this entropy coupling matrix, the system successfully predicted the abnormal blood pressure fluctuations that might occur within 15-20 minutes after the patient's local temperature increased (0.8℃ / h relative to the baseline), enabling medical staff to promptly detect and treat bleeding at the puncture site and prevent hematoma formation.

[0048] In one embodiment, the abnormal physical sign analysis unit 1 further includes a path construction module 103 (not shown in the figure), which is used to establish an abnormal physical sign propagation path map based on the eigenvector decomposition result of the abnormal physical sign coupling matrix.

[0049] Specifically, the parameter generation module 101 is connected to the path construction module 103 through the matrix modeling module 102.

[0050] Specifically, taking the monitoring data of a 65-year-old female patient in the above embodiment as an example, the abnormal sign coupling matrix M obtained by the matrix modeling module 102 is decomposed into eigenvectors.

[0051] Specifically, first, the eigenvalues ​​and eigenvectors of the abnormal physical sign coupling matrix M are calculated: λ1 = 1.85 (largest eigenvalue); v1=[0.62,0.45,0.31,-0.28,0.33,0.35] (corresponding feature vector); λ² = 1.23; v2=[0.41,0.38,-0.35,0.42,0.39,-0.33]; ... (other eigenvalues ​​and eigenvectors).

[0052] Specifically, then, based on the eigenvector v1 corresponding to the largest eigenvalue, the main propagation path is determined: Local temperature change rate (0.62) → blood pressure fluctuation amplitude (-0.28): strongest transmission pathway; Tissue tension index (0.45) → Peripheral vascular resistance index (0.35): Secondary propagation pathway; Blood leakage measurement (0.31) → Heart rate variability (0.33): Third transmission pathway.

[0053] Specifically, finally, based on the sign and magnitude of the feature vector components, an abnormal sign propagation path map G is constructed, including: The node set V: {l1,l2,l3,g1,g2,g3} represents six vital signs parameters; Edge set E: {(l1→g1),(l2→g3),(l3→g2)} represents the main propagation path; Weight W: The weight of an edge is determined based on the absolute value of its eigenvector components; The propagation pathway map of these abnormal signs shows that local temperature abnormalities are the earliest warning signal, spreading to overall signs through blood pressure fluctuations; changes in tissue tension mainly affect peripheral vascular resistance; and there is an interaction between bleeding volume and heart rate variability. This provides a clear order and focus for subsequent clinical monitoring.

[0054] In one embodiment, the sensitivity extraction unit 2 includes: The phase division module 201 (not shown in the figure) is used to extract the phase feature change sequence based on the temporal characteristics of the abnormal signs propagation path map, and determine the time boundaries of the acute phase, stable phase and recovery phase through dynamic time warping and threshold segmentation. Sensitivity calculation module 202 (not shown in the figure) is used to divide the monitoring stages based on time boundaries, analyze the correlation strength between the vital signs parameters of each monitoring stage, and determine the sensitivity coefficient of the monitoring indicators of each stage. The feature set generation module 203 (not shown in the figure) is used to perform time-frequency analysis on the sensitivity coefficients of monitoring indicators at each stage using a piecewise wavelet transform algorithm, and generate a sensitivity change feature set.

[0055] Specifically, the stage division module 201 is connected to the feature set generation module 203 through the sensitivity calculation module 202.

[0056] In one embodiment, the stage segmentation module 201, when extracting the phase feature change sequence based on the temporal characteristics of the abnormal sign propagation path map and determining the time boundaries of the acute phase, stable phase, and recovery phase through dynamic time warping and threshold segmentation, includes: S211. Extract the time series data of local-to-global sign parameter pairs from the abnormal sign propagation path map, and calculate the phase feature change sequence between parameter pairs; Specifically, based on the abnormal sign propagation path map G obtained by the path construction module 103, the monitoring data of the 65-year-old female patient were further analyzed; time series data were extracted from the three main propagation paths (l1→g1), (l2→g3) and (l3→g2), with a total monitoring duration of T=24h and a sampling interval of 1 minute, resulting in 1440 sampling points.

[0057] Specifically, phase features are extracted for the parameter pairs of each propagation path. Taking the strongest propagation path (l1→g1) as an example, the original data is first normalized to eliminate the influence of dimensions: l1'(t)=(l1(t)-min(l1)) / (max(l1)-min(l1)); g1'(t)=(g1(t)-min(g1)) / (max(g1)-min(g1)).

[0058] Specifically, the phase characteristic change sequence φ(t) of the normalized data is calculated: φ(t)=2π·[l1'(t)-g1'(t)] / [max(l1'-g1')-min(l1'-g1')]; The obtained φ(t) reflects the evolution of the phase difference between the local temperature change rate and the blood pressure fluctuation amplitude.

[0059] S212. Based on the phase feature change sequence, a dynamic time warping algorithm is used to align the local-to-global vital signs parameters in time and generate a monitoring time series change curve. Specifically, the distance matrix for the Dynamic Time Warping (DTW) algorithm is constructed. First, a local distance metric is defined: d(i,j)=|φ(i)-φ(j)|+λ·|ij|; In the formula, λ=0.05 is the time penalty factor, which is used to control the flexibility of time alignment.

[0060] Specifically, the cumulative distance matrix D is calculated recursively: D(1,1)=d(1,1); D(i,1)=D(i-1,1)+d(i,1), i>1; D(1,j)=D(1,j-1)+d(1,j), j>1; D(i,j)=d(i,j)+min{D(i-1,j),D(i-1,j-1),D(i,j-1)},i,j>1.

[0061] Specifically, based on the cumulative distance matrix D, a dynamic programming algorithm is used to solve for the optimal regularized path W. Tracing backwards from the endpoint (N,M) to the starting point (1,1), the time-aligned path is obtained: W={w1,w2,...,w K}, wk=(i k ,j k ); In the formula, the path length K=1562, which is greater than the original sequence length 1440, indicating that there is a certain time scaling.

[0062] Specifically, based on the optimal normalized path W, the time-aligned monitoring time series change curve C(t) is generated: C(t) = |φ(i k )-φ(j k )| / K, k=1,2,...,K.

[0063] S213. Perform double-threshold iterative segmentation on the monitored time-series change curves to determine the time boundary points of the acute phase, stable phase, and recovery phase; specifically including: S2131. Based on the characteristics of the monitoring time-series change curve, calculate the initial dual thresholds and preliminarily divide the curve into three stages: acute phase, stable phase, and recovery phase. S2132. Calculate the mean of the three stages, update the double threshold, and repeat the iterative calculation until the double threshold converges. S2133. Based on the converged double threshold, determine the time boundary points of the three stages and formulate the corresponding monitoring frequencies.

[0064] Specifically, the monitoring of time-series change curve characteristics includes the curve mean and standard deviation. This invention employs the Otsu method for dual-threshold iterative segmentation. Initialization parameters: μ=mean(C(t))=0.42 (curve mean); σ = std(C(t)) = 0.15 (standard deviation); Initial threshold: θ1=μ+σ=0.57, θ2=μ-σ=0.27.

[0065] Specifically, the iterative optimization process: First iteration: Acute phase interval: C(t) > 0.57, μ1 = 0.68; The steady-state interval is: 0.27 ≤ C(t) ≤ 0.57, μ2 = 0.42; Recovery period interval: C(t) < 0.27, μ3 = 0.21; Update threshold: θ1 = (0.68 + 0.42) / 2 = 0.55, θ2 = (0.42 + 0.21) / 2 = 0.32; Second iteration: Acute phase interval: C(t)>0.55, μ1=0.67; The steady-state interval is: 0.32 ≤ C(t) ≤ 0.55, μ2 = 0.43; Recovery period interval: C(t) < 0.32, μ3 = 0.22; Update threshold: θ1 = 0.55, θ2 = 0.33. After 5 iterations, the thresholds converged to θ1=0.54 and θ2=0.33.

[0066] Specifically, the final determined time boundary points are: t1=2.5h (150 minutes post-surgery): the dividing point between the acute and stable phases, at which point C(t) first stabilizes below θ1; t2=16h (960 minutes post-surgery): the dividing point between the stable period and the recovery period, at which point C(t) remains below θ2 for 30 minutes. The phased results show: Acute phase (0-2.5h): The phase characteristics change drastically, C(t)>0.54, requiring monitoring every 5 minutes; Stable period (2.5-16h): Phase characteristic fluctuations slow down, 0.33≤C(t)≤0.54, monitoring can be performed every 15 minutes; Recovery period (16-24h): Phase characteristics are basically stable, C(t)<0.33, and can be monitored every 30 minutes; This data-driven, dynamic phasing method not only provides precise time points but also offers a quantitative basis for determining the monitoring frequency at different stages.

[0067] In one embodiment, when the sensitivity calculation module 202 divides the monitoring stages based on time boundaries and analyzes the correlation strength between the vital sign parameter pairs in each monitoring stage to determine the sensitivity coefficient of the monitoring indicators in each stage, it includes: S221. Based on the time boundaries of each monitoring stage, extract the time series dataset of local-global vital signs parameters at different stages; Specifically, based on the time boundary points t1=2.5h and t2=16h obtained by the stage division module 201, the monitoring data of this 65-year-old female patient was divided into three stages: Acute phase: T1 = [0, 2.5 h]; Steady-state period: T2 = [2.5h, 16h]; Recovery period: T3 = [16h, 24h].

[0068] Specifically, parameter pairs of data for the three main propagation paths are extracted for each time period: D1={(l1,g1),(l2,g3),(l3,g2)}|T1 / / Acute phase dataset; D2={(l1,g1),(l2,g3),(l3,g2)}|T2 / / Stability period dataset; D3={(l1,g1),(l2,g3),(l3,g2)}|T3 / / Rehabilitation dataset.

[0069] Specifically, taking the strongest propagation path (l1, g1) as an example, the number of sampling points in each stage are as follows: n1=150 (acute phase, 5 minutes / sampling); n2=810 (stable period, 15 minutes / sampling); n3=480 (Recovery period, 30 minutes / sampling).

[0070] S222. Calculate the correlation strength between parameter pairs at each stage and construct a synchronization index matrix to quantify the degree of correlation between vital signs parameters at different stages. Specifically, the generalized synchronization index (GSI) method is used to calculate the correlation strength of parameter pairs. For a parameter pair (x, y), its GSI is defined as: GSI(x,y)=1-<|x(t+τ)-F(y(t))|>t / σ x ; In the formula, F(·) is the nonlinear mapping function, τ is the time delay, and σ is the time delay. x Let t be the standard deviation of parameter x, and t represent the time average.

[0071] Specifically, taking the acute phase T1 as an example, the 3×3 synchronization index matrix S1 is calculated: S1(l1,g1)=0.85S1(l1,g2)=0.42S1(l1,g3)=0.38; S1(l2,g1)=0.45S1(l2,g2)=0.39S1(l2,g3)=0.72; S1(l3,g1)=0.41S1(l3,g2)=0.68S1(l3,g3)=0.44.

[0072] Specifically, the synchronization index matrix S2 for the stable period T2 is calculated similarly: S2(l1,g1)=0.62S2(l1,g2)=0.35S2(l1,g3)=0.31; S2(l2,g1)=0.38S2(l2,g2)=0.33S2(l2,g3)=0.58; S2(l3,g1)=0.34S2(l3,g2)=0.55S2(l3,g3)=0.36.

[0073] Synchronization index matrix S3 for T3 during the recovery period: S3(l1,g1)=0.45S3(l1,g2)=0.28S3(l1,g3)=0.25; S3(l2,g1)=0.31S3(l2,g2)=0.27S3(l2,g3)=0.42; S3(l3,g1)=0.29S3(l3,g2)=0.41S3(l3,g3)=0.30.

[0074] S223. Based on the correlation between parameters in the synchronization index matrix, calculate the sensitivity coefficient of each monitoring indicator relative to the overall change in physical characteristics.

[0075] Specifically, define local trait parameters l i Sensitivity coefficient relative to overall physical characteristics: η(l i )=Σj[w(j)·S(l i ,g j )]; In the formula, w(j) represents the overall vital sign parameter g. j The weighting coefficients are set based on expert experience: w(1)=0.5, w(2)=0.3, w(3)=0.2.

[0076] Specifically, the sensitivity coefficients of various local signs and parameters during the acute phase T1 are calculated: η1(l1)=0.85×0.5+0.42×0.3+0.38×0.2=0.62; η1(l2)=0.45×0.5+0.39×0.3+0.72×0.2=0.48; η1(l3)=0.41×0.5+0.68×0.3+0.44×0.2=0.49.

[0077] Specifically, the sensitivity coefficient for the stable period T2 is calculated similarly: eta2(l1)=0.47, eta2(l2)=0.41, eta2(l3)=0.40; Sensitivity coefficient of T3 during the recovery period: eta3(l1)=0.35, eta3(l2)=0.32, eta3(l3)=0.32.

[0078] In one embodiment, the feature set generation module 203 uses a piecewise wavelet transform algorithm to perform time-frequency analysis on the sensitivity coefficients of monitoring indicators at each stage, and generates a sensitivity change feature set.

[0079] Specifically, in this embodiment, Morlet wavelet is used to perform time-frequency analysis on the sensitivity coefficient sequence. For each monitoring stage k, a feature vector is constructed: F k =[η k (l1),η k (l2),η k (l3),δ k (l1),δ k (l2),δ k (l3)]; In the formula, δ k (l i ) represents the parameter l for the k-th stage. i The rate of change in sensitivity.

[0080] Specifically, the feature vectors for the three stages are calculated: F1=[0.62,0.48,0.49,-0.15,-0.07,-0.09] / / Acute phase; F2=[0.47,0.41,0.40,-0.12,-0.09,-0.08] / / Stability period; F3=[0.35,0.32,0.32,-0.06,-0.04,-0.04] / / Rehabilitation period.

[0081] Specifically, these feature vectors are combined into a sensitivity change feature set Ω={F1,F2,F3}, which is used for subsequent early warning signal generation.

[0082] In one embodiment, the early warning signal generation unit 3 generates graded early warning signals and intervention suggestions based on the sensitivity change feature set and the comparison results of risk probability and preset threshold.

[0083] Specifically, the early warning signal generation unit 3 includes: Risk assessment module 301 (not shown in the figure) is used to calculate the risk probability based on the sensitivity change feature set; The signal generation module 302 (not shown in the figure) is used to determine the warning level based on the comparison result between the risk probability and the preset threshold. It is recommended that module 303 (not shown in the figure) be used to generate corresponding intervention suggestions based on the warning level.

[0084] Specifically, the risk assessment module 301 employs a support vector machine (SVM) algorithm to train a risk probability model based on historical case data. For the feature vector F... k Calculate its risk probability: P(F k )=SVM(F k )∈[0,1].

[0085] Specifically, a three-level early warning threshold is set: θ H =0.8 (high-risk threshold); θ M =0.5 (medium risk threshold); θ L =0.2 (low risk threshold).

[0086] Specifically, an early warning signal is generated based on the comparison between the risk probability and the threshold: When P(F) k )≥θ H At that time, a red alert (Level I) will be issued; When θ M ≤P(F k )<θ H At that time, a yellow alert (Level II) will be issued; When θ L ≤P(F k )<θ M At that time, a blue alert (Level III) will be issued. When P(F) k )<θ L At that time, the system was in normal monitoring status.

[0087] Specifically, regarding the monitoring data for this patient: Acute phase: P(F1) = 0.85 > θ H This triggered a red alert. Stable period: P(F2) = 0.45 < θ M This triggered a blue alert. Recovery period: P(F3) = 0.15 < θ L The system is in normal monitoring status.

[0088] To facilitate understanding of the above technical solution of the present invention, the following is a detailed explanation using a 65-year-old female patient in the intensive care unit of a hospital as an example: This patient underwent percutaneous coronary intervention (PCI) for coronary artery disease and required 24-hour post-procedure monitoring. The system successfully captured the following key events during the monitoring process: Two hours post-procedure, the system detected a sudden increase in the sensitivity coefficient of the local temperature change rate (l1) to 0.62, triggering a red alert. Medical staff promptly examined the site and found minor bleeding at the puncture point; the situation was brought under control after timely compression. Eight hours after the operation, the system showed that the overall sensitivity coefficient dropped to a stable level (about 0.4), and the system entered a blue alert state. The medical staff adjusted the monitoring frequency accordingly. Twenty hours after the surgery, all indicators were within the normal range, the system exited the warning state, and the patient successfully completed postoperative monitoring.

[0089] The above cases verify the effectiveness and practical value of the present invention in clinical practice. Through multi-level data analysis and dynamic early warning mechanisms, the system provides medical staff with accurate monitoring data and timely intervention suggestions, effectively preventing the occurrence of postoperative complications.

[0090] In summary, by utilizing the technical solutions described above, this invention analyzes the correlation between local and global vital signs through the construction of a cross-entropy coupling model, generates information transmission volume and transmission directionality index using the cyclic correlation entropy algorithm, and establishes an abnormal vital sign coupling matrix by combining generalized fractal dimension calculation, thereby achieving in-depth mining of vital sign data of patients after femoral artery puncture. Simultaneously, this invention can accurately capture the dynamic correlation between local vital sign parameters (puncture point temperature change rate, tissue tension index, and bleeding volume measurement) and global vital sign parameters (blood pressure fluctuation amplitude, heart rate variability, and peripheral vascular resistance index). Furthermore, it constructs an abnormal vital sign propagation path map through matrix eigenvector decomposition, providing a reliable data foundation for subsequent monitoring and early warning. This multi-dimensional information fusion-based analysis method significantly improves the ability to identify early signs of postoperative complications. This invention designs a complete sensitivity extraction mechanism. It uses a dynamic time warping algorithm to align local-to-global vital signs parameters over time, employs a dual-threshold iterative segmentation method to determine the time boundaries of the acute, stable, and recovery phases, and utilizes a generalized synchronization index to calculate the correlation strength between parameter pairs at each stage. Through a phased sensitivity analysis method, this invention not only achieves precise staging of monitoring data but also quantifies the correlation between vital signs parameters at different stages by constructing a synchronization index matrix, thereby calculating the sensitivity coefficient of each monitoring indicator relative to changes in overall vital signs. Furthermore, through time-frequency analysis using a piecewise wavelet transform algorithm, the generated sensitivity change feature set comprehensively reflects the characteristics of vital sign changes in patients at different recovery stages, providing a more accurate quantitative basis for clinical monitoring. This invention establishes a scientific early warning mechanism based on a sensitivity change feature set. It trains a risk probability model using a support vector machine algorithm and combines this with preset multi-level early warning thresholds to achieve automatic grading of early warning signals. Simultaneously, the system can promptly issue early warning signals of different levels based on the calculated risk probability and generate corresponding intervention suggestions. This data-driven dynamic early warning mechanism not only improves the accuracy of early warnings for postoperative complications but also achieves rational allocation of monitoring resources by adaptively adjusting monitoring frequency, providing scientific decision support for medical staff. Furthermore, the system's tiered early warning function can effectively reduce false alarm rates, avoid excessive consumption of medical resources, and improve the overall efficiency of clinical monitoring.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A data processing system for monitoring vital signs after femoral artery puncture, characterized in that, include: The abnormal signs analysis unit is used to construct a cross-entropy coupling model based on the signs data of patients after femoral artery puncture, in order to analyze the correlation between local signs and global signs and obtain the abnormal signs propagation path map. The sensitivity extraction unit is used to divide different monitoring stages and perform phase feature analysis based on the abnormal vital signs propagation path map, extract sensitivity coefficients, and generate a sensitivity change feature set; The early warning signal generation unit is used to generate graded early warning signals and intervention suggestions based on the sensitivity change feature set and the comparison results of risk probability and preset threshold. The abnormal signs analysis unit includes: The matrix modeling module uses the cyclic correlation entropy algorithm to generate information transmission volume and transmission directionality index, and establishes anomaly characteristic coupling matrix; Among these methods, the information transmission volume and transmission directionality index are generated using the cyclic correlation entropy algorithm, and an anomaly characteristic coupling matrix is ​​established, including: Based on the local and global symptom parameter sets, the generalized fractal dimension is calculated, and the information transmission volume is generated by combining the cyclic correlation entropy algorithm. Based on the amount of information transmitted, analyze the phase synchronization and information flow direction between parameters to generate a transmission directionality index; Based on the information transmission volume and transmission directionality index, a coupling strength matrix and a polarity matrix are constructed, and an abnormal symptom coupling matrix is ​​generated through matrix operations.

2. The femoral artery puncture postoperative vital sign monitoring data processing system according to claim 1, characterized in that, The abnormal vital signs analysis unit is connected to the early warning signal generation unit through the sensitivity extraction unit; The abnormal vital signs analysis unit also includes: The parameter generation module is used to establish local and global vital sign parameter sets based on the vital sign data of patients after femoral artery puncture, and to construct a cross-entropy coupling model. The path construction module is used to build a path map of abnormal signs based on the eigenvector decomposition results of the abnormal sign coupling matrix.

3. The data processing system for monitoring vital signs after femoral artery puncture according to claim 2, characterized in that, The set of local vital signs parameters includes the puncture site temperature change rate, tissue tension index, and bleeding volume measurement value. The overall set of vital signs parameters includes blood pressure fluctuation amplitude, heart rate variability, and peripheral vascular resistance index; The expression for the cross-entropy coupling model is: ; ; In the formula, H ( L , G ) represents the cross-entropy from the local layer to the global layer of the model; P ( l i , g j ) represents the first local symptom parameter set. i The first parameter and overall vital signs parameter set j The joint probability distribution of the parameters; P ( g j | l i () indicates that the local vital signs parameters are known. l i Overall vital signs parameters under certain conditions g j The conditional probability; C This refers to the coupling coefficient from the local layer to the global layer of the model. H max This is the theoretical maximum cross-entropy value.

4. The data processing system for monitoring vital signs after femoral artery puncture according to claim 1, characterized in that, The step of calculating the generalized fractal dimension based on the local and global characteristic parameter sets, and generating the information transmission volume by combining it with the cyclic correlation entropy algorithm, includes: Based on a fixed time interval, the parameter sequences in the local and global vital signs parameter sets are divided into several subsequences. The local variance of each subsequence is calculated to construct a scaling transformation function. The Hearst exponent of each parameter time series is obtained by linear fitting under logarithmic coordinates. Based on the Hearst exponent, a generalized fractal dimension and quality index are constructed. By calculating the spectral width and peak position of the multifractal spectrum, the complexity of each parameter sequence and the dominant fractal characteristics are quantified. The information entropy of each parameter sequence is calculated based on the generalized fractal dimension, and then weighted and fused using the cyclic correlation entropy algorithm to generate the information transmission amount of local vital signs parameters to overall vital signs parameters.

5. The femoral artery puncture postoperative vital sign monitoring data processing system according to claim 4, characterized in that, The expression for the amount of information transmitted is: ; ; ; ; In the formula, Q ( l i , g j ) represents local vital signs parameters l i With overall vital signs parameters g j The amount of information transmitted between them; w 1 represents the information entropy weighting coefficient; w 2 represents the weighting coefficient of the cyclic correlation entropy; w 3 indicates the spectral width weighting coefficient; S ( q ) represents the information entropy of vital signs parameters; D q The generalized fractal dimension representing vital signs parameters; N This indicates the number of sampling points for the monitoring data; CE ( l i , g j ) represents local vital signs parameters l i With overall vital signs parameters g j Cyclic related entropy; Δ α The spectral width represents the vital signs parameter; I ( l i ) represents local vital signs parameters l i The amount of information; I ( g j ) represents overall vital signs parameters g j The amount of information; I ( l i , g j ) indicates parameter pair ( l i , g j The amount of joint information; R ( t ) indicates based on observation time t Regulatory factors; Δ α The spectral width represents the vital signs parameter; α max This represents the maximum value of the singularity index of the vital signs parameter; α min This represents the minimum value of the singularity index of the vital signs parameter.

6. The data processing system for monitoring vital signs after femoral artery puncture according to claim 1, characterized in that, The step of generating a transmission directionality index by analyzing the phase synchronization and information flow direction between parameters based on the amount of information transmitted includes: The parameters in the local and global characteristic parameter sets are paired up, and the Hilbert transform is performed on each pair of parameter sequences to obtain the corresponding analytical signals. The phase difference sequence of each pair of parameters is calculated and the probability distribution is statistically analyzed. The phase synchronization index is obtained by normalization to characterize the synchronization coupling strength of the parameter pair. The bidirectional information flow for each pair of parameters is calculated based on the transfer entropy, and the dominant direction of information transmission is determined by the directional index, thus obtaining the transmission directionality index.

7. The data processing system for monitoring vital signs after femoral artery puncture according to claim 1, characterized in that, The sensitivity extraction unit includes: The phase division module is used to extract the phase feature change sequence based on the temporal characteristics of the abnormal sign propagation path map, and determine the time boundaries of the acute phase, stable phase and recovery phase through dynamic time warping and threshold segmentation. The sensitivity calculation module is used to divide the monitoring stages based on time boundaries, analyze the correlation strength between the vital signs parameters in each monitoring stage, and determine the sensitivity coefficient of the monitoring indicators in each stage. The feature set generation module is used to perform time-frequency analysis on the sensitivity coefficients of monitoring indicators at each stage using a piecewise wavelet transform algorithm, and generate a sensitivity change feature set.

8. A data processing system for monitoring vital signs after femoral artery puncture according to claim 7, characterized in that, The stage segmentation module, when extracting phase feature change sequences based on the temporal characteristics of the abnormal sign propagation path map and determining the time boundaries of the acute, stable, and recovery phases through dynamic time warping and threshold segmentation, includes: Extract time series data of local-to-global sign parameter pairs from the abnormal sign propagation path map, and calculate the phase feature change sequence between parameter pairs; Based on the phase feature change sequence, a dynamic time warping algorithm is used to align the local-to-global vital signs parameters in time and generate monitoring time-series change curves. The monitoring time-series change curves are segmented using a dual-threshold iterative method to determine the time boundary points of the acute phase, stable phase, and recovery phase.

9. A data processing system for monitoring vital signs after femoral artery puncture according to claim 7, characterized in that, The sensitivity calculation module, when dividing the monitoring stages based on time boundaries and analyzing the correlation strength between pairs of vital signs parameters in each monitoring stage to determine the sensitivity coefficient of the monitoring indicators in each stage, includes: Based on the time boundaries of each monitoring stage, extract local-to-global vital signs parameter pairs from time-series datasets at different stages; The correlation strength between parameter pairs at each stage was calculated, and a synchronization index matrix was constructed to quantify the degree of correlation between vital signs parameters at different stages. Based on the correlation between parameters in the synchronization index matrix, the sensitivity coefficient of each monitoring indicator relative to the overall change in vital signs is calculated.

10. A data processing system for monitoring vital signs after femoral artery puncture according to claim 8, characterized in that, The step of performing dual-threshold iterative segmentation on the monitored time-series change curves to determine the time boundary points of the acute phase, stable phase, and recovery phase includes: Based on the characteristics of the monitoring time-series change curve, an initial dual threshold is calculated, and the curve is initially divided into three stages: acute phase, stable phase, and recovery phase. Calculate the mean of the three stages, update the double threshold, and repeat the iterative calculation until the double threshold converges; Based on the converged double thresholds, the time boundary points of the three stages are determined, and the corresponding monitoring frequencies are established.

Citation Information

Patent Citations

  • Intensive care data acquisition and early warning system

    CN119423700A

  • Intelligent nursing management system and method for critical patients

    CN119453949A

  • Intelligent early warning system for lower limb ischemia based on multi-parameter monitoring

    CN119867682A

  • Hemodialysis care unit extracorporeal circulator monitoring method and system and storage medium

    CN119993463A

  • Vascular surgery intervention intracavity treatment postoperative complication prediction and management system

    CN120148883A