Cardiovascular data monitoring method for cardiovascular medicine department

Abstract

The invention relates to the technical field of cardiovascular monitoring, in particular to a cardiovascular data monitoring method for the cardiovascular medicine department, which comprises the following steps of: performing sparse regularization processing on characteristic parameters of blood pressure waveform and pulse wave velocity based on a variable set of cardiovascular high-dimensional data; according to the method, redundant variables in high-dimensional data are optimized through sparse regularization processing, a dependency matrix is constructed to extract weight distribution, the accuracy and efficiency of data processing are improved, deep integration of multi-source data is achieved based on variable relevance optimization and distribution parameter updating, the fine analysis capacity of dynamic signals is remarkably enhanced, and the method is suitable for large-scale popularization and application. Through time sequence feature extraction and anomaly labeling, accurate positioning of an abnormal interval is improved, the distribution rule analysis effect of electrocardiosignals is enhanced, dynamic change rate calculation and fluctuation error correction of the pulse wave speed are combined, the reliability of dynamic trend prediction is guaranteed, extraction and evaluation of key risk factors are further enhanced, and the reliability of dynamic trend prediction is improved. And dynamic assessment and accurate quantification of cardiovascular risks are realized.

Description

Technical Field

[0001] The present invention relates to the field of cardiovascular monitoring technology, and in particular to a cardiovascular data monitoring method in cardiovascular medicine. Background Art

[0002] The field of cardiovascular monitoring technology includes technologies for measuring, recording and analyzing physiological parameters of the cardiovascular system, which are mainly used for the diagnosis, evaluation and monitoring of cardiovascular diseases. The core content of this technical field includes ECG signal monitoring, blood pressure measurement, heart rate variability analysis and other physiological parameter collection and processing methods that reflect the state of cardiovascular function. The overall technical field systematically covers the design of data acquisition devices, the research of physiological parameter measurement methods and the development of related data transmission and storage technologies, and also involves the optimization of data analysis methods to improve the accuracy and reliability of monitoring.

[0003] Among them, the cardiovascular data monitoring method of the Department of Cardiology refers to the method of collecting and recording physiological parameters for patients with cardiovascular diseases. The subject of this patent covers the collection of ECG signals and hemodynamic data through precise measurement equipment, and the extraction of characteristic parameters in combination with specific signal processing methods, and the use of data screening and integration methods to organize and analyze multi-source physiological data, thereby achieving comprehensive monitoring of cardiovascular function. This method generally uses a dynamic data acquisition device for long-term monitoring, uses high-precision sensors and calibration methods to ensure the accuracy of the data, and combines specific data analysis methods to achieve the monitoring goals.

[0004] Existing technologies have difficulty in effectively eliminating redundant variables when processing high-dimensional physiological data, resulting in analysis results being affected by noise. The analysis of variable correlation lacks a dynamic optimization mechanism, and the ability to integrate multi-source data is insufficient, making it difficult to fully reflect the cardiovascular function status. The ability to locate abnormal intervals is weak, and the potential feature distribution in time series signals cannot be accurately identified. The analysis accuracy of nonlinear dynamic change rates and trend characteristics is low. Cardiovascular risk assessment is limited to static indicator analysis and fails to reflect dynamic risk change trends, affecting the comprehensive monitoring and early warning effects of the disease. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a cardiovascular data monitoring method for cardiovascular medicine.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a cardiovascular data monitoring method of cardiovascular medicine, comprising the following steps: S1: Based on the variable set of cardiovascular high-dimensional data, the characteristic parameters of blood pressure waveform and pulse wave velocity are sparsely regularized. The linear relationship is analyzed variable by variable, weak correlation edges are screened in descending order, redundant variables are eliminated and a dependency matrix is ​​established. The weight distribution is extracted from the relationship matrix to generate sparse dependency weight values. S2: Based on the sparse dependency weight value, the relationship between heart rate variability and arteriosclerosis index is distributedly fitted, the deviation amount and distribution parameters of multiple variables are updated, the strong correlation between variables is optimized, the relationship matrix is ​​reconstructed, and the structural risk weight value matrix is ​​generated; S3: Based on the structural risk weight value matrix, extract feature changes from the time series data of the electrocardiogram signal in segments, encode the features and mark the abnormal intervals, analyze the time series changes in combination with the characteristics of the encoding values, and establish the time series potential distribution encoding value; S4: Based on the time series potential distribution coding value, the dynamic change rate of the implicit state is calculated using the coding input of the pulse wave velocity, the nonlinear offset characteristics are analyzed, the trend change values ​​are merged through hierarchical modeling, the fluctuation error in the parameter set is corrected, and the time series dynamic trend prediction value is obtained; S5: Based on the time series dynamic trend prediction value and the structural risk weight value matrix, cross-analysis of blood pressure waveform and heart rate variability parameters is performed to extract the dynamic change trend of key risk factors, and the joint evaluation index is calibrated to generate a joint index value of cardiovascular risk trend, implement individualized risk warning, optimization of intervention measures, and remote monitoring management.

[0007] Based on the combined index values ​​of the cardiovascular risk trend, individualized risk warning is divided into short-term, medium-term and long-term risk assessments. Short-term risk assessment calculates risk weights based on the time domain and frequency domain parameters of blood pressure fluctuation range and heart rate variability (HRV) indicators, such as SDNN (standard deviation), RMSSD (root mean square difference between adjacent normal heartbeat intervals), LF / HF (low frequency / high frequency ratio), etc. The current individual risk level is determined by setting a threshold interval (such as SDNN < 50 ms for high risk, 50-100 ms for medium risk, and >100 ms for low risk). Medium-term risk assessment combines parameters such as pulse wave velocity (PWV) and arterial compliance index (CAI) to calculate the degree of vascular sclerosis. The formula PWV = Δx / Δt (where Δx is the arterial path length and Δt is the pulse propagation time) is used in combination with the age-adjusted assessment criteria (such as PWV > 10 m / s for high risk, 8-10 m / s for medium risk, and <8 m / s for low risk). For long-term risk assessment, the incidence of individual cardiovascular events is statistically analyzed, historical data trends are analyzed in combination with electronic health records (EHR), and a multivariate regression model is established to calculate the probability of future risk. Intervention measures are optimized by adjusting lifestyle and drug intervention plans based on individual risk levels. For those with abnormal HRV, it is recommended to increase cardiopulmonary training (such as 30 minutes of moderate-intensity exercise 3 times a week), and for those with higher PWV, antihypertensive drug intervention (such as calcium channel blockers, ACEI drugs) is recommended. Remote monitoring and management uses wearable devices to upload cardiovascular parameters in real time, and identifies abnormal patterns through machine learning algorithms. For example, LSTM (long short-term memory network) is used to predict the trend of cardiovascular parameter changes in the next cycle. Long-term health assessment uses dynamic data normalization methods (such as Z-score normalization) to compare historical trends. By setting the mean μ and standard deviation σ, the standardized value Z = (X - μ) / σ is calculated and compared with the baseline data of the population. If Z > 2 is considered a significant abnormality, telemedicine intervention is triggered, and finally a long-term individual cardiovascular health assessment result is formed.

[0008] The sparse dependency weight value includes weight distribution, sparse characteristics, and dependency matrix. The structural risk weight value matrix includes strong correlation between variables, deviation distribution parameters, and optimized relationship matrix. The time series potential distribution coding value includes characteristic change amount, abnormal interval annotation value, and coding value characteristic analysis result. The time series dynamic trend prediction value includes dynamic change rate, nonlinear offset characteristics, and trend change value. The cardiovascular risk trend joint indicator value includes dynamic change trend, key risk factors, and joint evaluation indicator correction results.

[0009] As a further solution of the present invention, the step of obtaining the sparse dependency weight value is specifically: S101: Based on the variable set of cardiovascular high-dimensional data, the initial characteristic parameters of blood pressure waveform and pulse wave velocity are extracted, the linear correlation between the variables and the waveform characteristics is analyzed one by one, the degree of dependence between the variables and the target characteristics is judged based on the correlation value, and the weakly correlated variables with correlation coefficients lower than the set threshold are screened to generate a weakly correlated variable set; S102: performing weighted calculation on multiple variables in the weakly associated variable set according to the distribution characteristics of their influence on the pulse wave velocity, calculating the redundancy value of each variable in the variable set, and eliminating variables with higher redundancy by comparing the redundancy value with a set threshold, thereby generating a variable set after eliminating redundancy; S103: In the variable set after removing redundancy, analyze the dependency relationship between variables, calculate the influence of variables on matrix weight distribution, construct a dependency matrix and extract weights, using the formula: ; Generate sparse dependency weight values; in, Representative variables and The dependency weight value of Representative variables and The dependency value of Representative variables and The distribution characteristic value of is the number of variables in the variable set, represents the variable being analyzed, As a mediating variable to measure the association, Another variable representing the dependency relationship.

[0010] As a further solution of the present invention, the steps of obtaining the structural risk weight value matrix are specifically as follows: S201: Based on the sparse dependency weight value, analyzing the relationship between heart rate variability and arteriosclerosis index, extracting variable parameters associated with the two, calculating regression coefficients of multiple parameters in a multivariate relationship, adjusting parameter ranges using fitting errors, and generating a preliminary relationship distribution model; S202: Based on the preliminary relationship distribution model, update the deviation of multiple parameters in the model, call the updated parameters to calculate their influence on the multivariate distribution, optimize the distribution parameter range through difference analysis, establish updated distribution characteristics, and generate an updated relationship distribution model; S203: using the updated relationship distribution model, calculating the dependency strength between variables, adjusting the distribution of strongly associated variables according to the influence of the variables on the weights in the relationship matrix, reconstructing the dependency matrix, and generating an optimized relationship matrix; S204: Based on the optimized relationship matrix, extract the influence weights of the variables in the matrix on the risk distribution, using the formula: ; Calculate and generate the structural risk weight value matrix; in, Representative variables and The structural risk weight value of Representative variables With variables The dependency value of Representative variables The actual impact weight of Representative variables and The theoretical impact distribution weight of is the total number of variables, and the summation symbol indicates the calculation of the cumulative risk value of all dependencies between variables.

[0011] As a further solution of the present invention, the step of obtaining the temporal potential distribution coding value is specifically as follows: S301: Based on the structural risk weight value matrix, extract the characteristic variation within a time period from the time series data of the electrocardiogram signal, analyze the abnormal intervals of the signal characteristic variation within multiple time periods, call the extracted characteristic variation and mark the abnormal fluctuation range through segmented comparison to generate characteristic variation data; S302: digitally encode the characteristic variation data, calculate the corresponding relationship between the encoding value and the change of the electrocardiogram time series waveform, generate the distribution characteristic data of each feature segment in combination with the marked abnormal interval, mark the abnormal encoding value range in the corresponding time period, and generate the encoded abnormal interval data; S303: combining the encoded abnormal interval data, calling the distribution comparison result between the abnormal coding value and the normal coding value, evaluating the time series fluctuation characteristics and establishing the distribution characteristics of each segment of data, and generating time series distribution characteristic data; S304: Based on the time series distribution characteristic data, a potential distribution model for each code value is established, and the distribution probability of the feature value in the differentiated time series interval is calculated using the formula: ; Generate time series latent distribution encoding values; in, Represents the coded value The probability density function of Representative The risk weight of the coded value, For the The average outlier of the coded values, For the The standard deviation of the coded values, is the total number of coded values, the exponential function is used to describe the distribution probability of the coded values, and the denominator is the normalization term for the normalization of the distribution.

[0012] As a further solution of the present invention, the step of obtaining the time series dynamic trend prediction value is specifically as follows: S401: Based on the data of the time series potential distribution code value, the time series data of the pulse wave velocity is coded and inputted, the dynamic change rate at each time point is calculated, the change trend is analyzed according to the time series fluctuation, and the dynamic change rate data is generated; S402: Analyze the nonlinear offset characteristics of the dynamic change rate data, combine the time series fluctuation trend results, perform segmented evaluation on the nonlinear offset value, optimize the data trend by adjusting the offset distribution, and generate nonlinear trend analysis data; S403: Integrate the trend change values ​​of the differentiated time points in combination with the nonlinear trend analysis data, adjust the multi-layer trend distribution based on hierarchical modeling, process the interaction between trend points, and generate merged trend change data; S404: Based on the merged trend change data, calculate the correction parameter centralized fluctuation error using the formula: ; Generate time series dynamic trend forecast values; in, Representative time The trend forecast value of is the overall correction factor, For the The trend weight of the time point, For the The adjustment factor for the time point, is the total number of time points, and the exponential function describes the nonlinear impact of trend changes at time points on the overall trend.

[0013] As a further solution of the present invention, the step of obtaining the cardiovascular risk trend combined index value is specifically: S501: using the time series dynamic trend prediction value and the structural risk weight value matrix, cross-analyzing the blood pressure waveform and the heart rate variability parameters, extracting the correlation of the fluctuation data and evaluating the key interaction points, and analyzing the parameter interaction trend changes to generate cross-analysis data; S502: extracting the dynamic change trend of key risk factors based on the cross-analysis data, evaluating the characteristic fluctuation of each risk factor in combination with the dynamic change curve, and generating key risk factor dynamic data by calculating the deviation rate and trend fitting value of key points; S503: combining the dynamic data of the key risk factors, calling the waveform analysis parameters and integrating the risk trends, and combining the deviation correction model of the assessment data, adjusting the weight values ​​by using the parameter distribution and optimizing the trend distribution results, and generating the corrected assessment model data; S504: Calculate the comprehensive change trend of risk factors in the time series using the corrected assessment model data, using the formula: ; Generate a combined index value for cardiovascular risk trend; in, Representative time The trend value of cardiovascular risk is the normalization coefficient, For the The weight of the risk factor, The peak risk time point is is the time diffusion parameter, is the total number of risk factors, and the exponential function describes the nonlinear impact of time changes in risk factors on the trend value.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, sparse regularization is used to optimize redundant variables in high-dimensional data, and a dependency matrix is ​​constructed to extract weight distribution, thereby improving the accuracy and efficiency of data processing. Based on variable correlation optimization and distribution parameter update, deep integration of multi-source data is achieved, and the fine analysis capability of dynamic signals is significantly enhanced. Through time series feature extraction and anomaly annotation, the accurate positioning of abnormal intervals is improved, and the analysis effect of the distribution law of electrocardiogram signals is enhanced. Combined with the calculation of the dynamic change rate of pulse wave velocity and the correction of fluctuation errors, the reliability of dynamic trend prediction is guaranteed, the extraction and evaluation of key risk factors are further strengthened, and the dynamic evaluation and accurate quantification of cardiovascular risks are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the main steps of the present invention; Figure 2 This is a flow chart of the steps for obtaining the sparse dependency weight value of the present invention; Figure 3 The following is a flow chart of the steps for obtaining the structural risk weight value matrix of the present invention; Figure 4 A flowchart of the steps for obtaining the temporal potential distribution coding value of the present invention; Figure 5 A flowchart of the steps for obtaining the time series dynamic trend prediction value of the present invention; Figure 6 This is a flow chart of the steps for obtaining the cardiovascular risk trend combined index value of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] Embodiment 1 See also Figure 1 The present invention provides a technical solution: a cardiovascular data monitoring method of cardiovascular medicine, comprising the following steps: S1: Based on the variable set of cardiovascular high-dimensional data, the characteristic parameters of blood pressure waveform and pulse wave velocity are sparsely regularized. The linear relationship is analyzed variable by variable, weak correlation edges are screened in descending order, redundant variables are eliminated and a dependency matrix is ​​established. The weight distribution is extracted from the relationship matrix to generate sparse dependency weight values. S2: Based on the sparse dependency weight value, the relationship between heart rate variability and arteriosclerosis index is fitted, the deviation and distribution parameters of multiple variables are updated, the strong correlation between variables is optimized, the relationship matrix is ​​reconstructed, and the structural risk weight value matrix is ​​generated; S3: Based on the structural risk weight value matrix, extract the feature changes from the time series data of the electrocardiogram signal in segments, encode the features and mark the abnormal intervals, analyze the time series changes based on the characteristics of the encoding value, and establish the time series potential distribution encoding value; S4: Based on the time series potential distribution coding value, the dynamic change rate of the implicit state is calculated using the coding input of the pulse wave velocity, the nonlinear offset characteristics are analyzed, the trend change values ​​are merged through hierarchical modeling, the fluctuation error in the parameter set is corrected, and the time series dynamic trend prediction value is obtained; S5: Based on the time series dynamic trend prediction value and the structural risk weight value matrix, cross-analysis of blood pressure waveform and heart rate variability parameters is performed to extract the dynamic change trend of key risk factors, and the joint evaluation index correction is performed to generate the cardiovascular risk trend joint index value, implement individualized risk warning, optimization of intervention measures, and remote monitoring management.

[0019] The sparse dependency weight values ​​include weight distribution, sparse characteristics, and dependency matrix. The structural risk weight value matrix includes strong correlation between variables, deviation distribution parameters, and optimized relationship matrix. The time series potential distribution coding values ​​include characteristic change amounts, abnormal interval annotation values, and coding value characteristic analysis results. The time series dynamic trend prediction values ​​include dynamic change rates, nonlinear offset characteristics, and trend change values. The cardiovascular risk trend joint indicator values ​​include dynamic change trends, key risk factors, and joint evaluation indicator correction results.

[0020] See also Figure 2 , the specific steps for obtaining the sparse dependency weight value are: S101: Based on the variable set of cardiovascular high-dimensional data, the initial characteristic parameters of blood pressure waveform and pulse wave velocity are extracted, the linear correlation between the variables and the waveform characteristics is analyzed one by one, the degree of dependence between the variables and the target characteristics is judged based on the correlation value, and the weakly correlated variables with correlation coefficients lower than the set threshold are screened to generate a weakly correlated variable set; The correlation coefficient is calculated based on the collected blood pressure waveform data and pulse wave velocity data, using the variable value change rate and average fluctuation amplitude as the calculation basis. The calculation formula is: ,in For variables and The covariance of and are the standard deviations of the variables respectively. By comparing the correlation coefficients between different variables, the variables below the threshold are screened out. The low threshold variables are based on the threshold interval. During the calculation process, the determined weakly correlated variable set removes the variables that do not meet the requirements one by one and saves the correlation records to form a weakly correlated variable set.

[0021] S102: performing weighted calculation on multiple variables in the weakly associated variable set according to the distribution characteristics of their influence on the pulse wave velocity, calculating the redundancy value of each variable in the variable set, and eliminating variables with higher redundancy by comparing the redundancy value with a set threshold, thereby generating a variable set after eliminating redundancy; Combined with the distribution range and waveform characteristics of each variable, the redundancy of the variable to the target feature is quantified, and the redundancy value is calculated using the formula: ,in is the redundancy value, is the distribution characteristic value of the variable, is the weight value between variables. By calculating the redundancy value of each variable and comparing it with the set threshold, the threshold refers to the fluctuation range , remove variables with high redundancy and generate a set of variables after eliminating redundancy.

[0022] S103: In the variable set after eliminating redundancy, analyze the dependency relationship between variables, calculate the influence of variables on the matrix weight distribution, construct the dependency matrix and extract the weight, using the formula: ; Generate sparse dependency weight values; in, Representative variables and The dependency weight value of Representative variables and The dependency value of Representative variables and The distribution characteristic value of is the number of variables in the variable set, represents the variable being analyzed, As a mediating variable to measure the association, Another variable representing the dependency relationship.

[0023] formula: ; The formula is useful in that by introducing the distribution characteristic value With dependency value The collaborative calculation quantifies the nonlinear dependencies between variables and normalizes the data, thus avoiding the weight imbalance problem caused by differences in data scale. Detailed explanation of the formula and the process of formula calculation and derivation: Representation variables and The dependency value of is obtained through covariance calculation during the collection process. The specific calculation method is: ; in, is a variable and The covariance of and are the standard deviations of the variables, respectively. The variable values ​​are calculated from the distribution data of the waveform characteristics; is a variable and The distribution characteristic value of the data after normalization is in the range of [0,1] and is calculated using the following formula: ; It is the weighted sum of variable dependency and feature distribution, reflecting the interaction strength between variables. Normalize the results to ensure that the range of the result value is suitable for the target feature; Calculation example: set up ,variable and The collection value is and , first calculate the covariance and standard deviation : ; ; for ,set up , after normalization, calculate: ; Substitute the weight formula into the calculation: ; ; This result shows that the weight value It indicates that there is a strong dependency between variables, which can be further used to optimize the calculation of sparse dependency weight values.

[0024] See also Figure 3 , the specific steps for obtaining the structural risk weight value matrix are: S201: Based on the sparse dependency weight value, the relationship between the heart rate variability and the arteriosclerosis index is analyzed, the variable parameters associated with the two are extracted, the regression coefficients of the multiple parameters in the multivariate relationship are calculated, the parameter range is adjusted using the fitting error, and a preliminary relationship distribution model is generated; First, the variable parameters are extracted from the collected heart rate variability and arteriosclerosis index data, and the correlation value between each variable and the target variable is calculated. The correlation coefficient formula is used to calculate Get the correlation between each variable and the target, where represents the covariance of variables x and y, and Represent the standard deviation of variables x and y respectively. Variables with lower correlation values ​​are selected for further fitting analysis. The regression coefficient of each variable is calculated by fitting the model and its parameter range is adjusted. When executing this process, according to the fitting error formula The model error is calculated and high error variables are excluded, and adjustments are made repeatedly until the error is minimized to generate a preliminary relationship distribution model.

[0025] S202: Based on the preliminary relationship distribution model, the deviation of multiple parameters in the model is updated, the updated parameters are called to calculate their influence on the multivariate distribution, the distribution parameter range is optimized through difference analysis, the updated distribution characteristics are established, and the updated relationship distribution model is generated; First, by analyzing the distribution characteristics of the variable parameters in the preliminary model, the mean and variance of the variable distribution are calculated, and the distribution parameters are adjusted to fit the data more accurately, using the normal distribution probability density function formula: , update the data distribution, where represents the mean of the data, Represents the standard deviation of the data, and then uses the optimized distribution characteristics to calculate the influence weight of each parameter on the multivariate distribution, through the weight analysis formula The variable weights are normalized, and finally an updated distribution characteristic model is established to generate an updated relationship distribution model.

[0026] S203: using the updated relationship distribution model, calculating the dependency strength between variables, adjusting the distribution of strongly associated variables according to the degree of influence of the variables on the weights in the relationship matrix, reconstructing the dependency matrix, and generating an optimized relationship matrix; First, the definition formula of the dependency matrix The strong correlation coefficient between variables is obtained, where Representation variables and The covariance of and Represents the variance of two variables, analyzes the impact of high-dependency coefficient variables on the weights in the relationship matrix, adjusts the matrix weights and recalculates the distribution characteristics between variables, and uses the normalization formula Reconstruct the dependency matrix and generate an optimized relationship matrix.

[0027] S204: Based on the optimized relationship matrix, extract the influence weight of the variables in the matrix on the risk distribution, using the formula: ; Calculate and generate the structural risk weight value matrix; in, Representative variables and The structural risk weight value of Representative variables With variables The dependency value of Representative variables The actual impact weight of Representative variables and The theoretical impact distribution weight of is the total number of variables, and the summation symbol indicates the calculation of the cumulative risk value of all dependencies between variables.

[0028] formula: ; The benefit of the formula is that it enhances the nonlinear relationship of risk distribution between variables by introducing a logarithmic function, while using weight accumulation calculation to effectively reduce the impact of single variable fluctuations on the overall structural risk weight value.

[0029] Detailed explanation of the formula and the process of formula calculation and derivation: set up , , , , , , , , , , then calculate: ; ; ; ; ; The result shows that the structural risk weight value between variables is 0.2775, indicating that the distribution weights of the variables in the current matrix have reached a medium level in terms of the balance of risk distribution. Further optimization can reduce the concentration of overall risk deviation.

[0030] See also Figure 4 , the specific steps for obtaining the temporal potential distribution encoding value are: S301: extracting feature changes within a time period from the time series data of the electrocardiogram signal based on the structural risk weight value matrix, analyzing abnormal intervals of signal feature changes within multiple time periods, calling the extracted feature changes and marking the abnormal fluctuation range through segmented comparison, and generating feature change data; First, the electrocardiogram signal is divided into equally spaced time windows. The characteristics of each time window are quantified by calculating the standard deviation, mean value and peak change rate. The main parameters of the signal change rate in the time window are extracted, and the fluctuation range of these parameters in the overall time series is calculated. The characteristics of abnormal points are identified by comparing the relative fluctuation levels of the parameter values ​​in each time window. The abnormal change points of each time window are recorded, and the corresponding feature changes are extracted. The feature changes of all windows are aggregated to form a feature change set to generate feature change data.

[0031] S302: digitally encode the feature variation data, calculate the corresponding relationship between the encoding value and the change of the electrocardiogram time series waveform, generate the distribution characteristic data of each feature segment in combination with the marked abnormal interval, mark the abnormal encoding value range in the corresponding time period, and generate the encoded abnormal interval data; By setting thresholds for the value range of each feature change, the feature change values ​​are classified into three intervals: low, medium, and high. The number of occurrences in each interval is calculated, and a unique coding value is assigned to each interval. For example, the low, medium, and high intervals correspond to coding values ​​0, 1, and 2, respectively. When marking abnormal intervals, the rapid changes in coding values ​​in adjacent windows are analyzed to detect significant points of signal fluctuations, identify the coding rules of abnormal intervals, count and record the time series distribution characteristics of each abnormal coding, and generate encoded abnormal interval data.

[0032] S303: combining the encoded abnormal interval data, calling the distribution comparison result between the abnormal coding value and the normal coding value, evaluating the time series fluctuation characteristics and establishing the distribution characteristics of each segment of data, and generating time series distribution characteristic data; Count the frequencies of different coding values ​​in the time series, calculate the relative change ratios between abnormal coding values ​​and normal coding values, quantify the characteristics of signal fluctuations by comparing these change ratios, evaluate the overall characteristics of each segment of data by calculating the weighted distribution average of each coding value, construct the time series distribution characteristics based on the results, and record the main distribution characteristics of each segment of coding data to generate time series distribution characteristic data.

[0033] S304: Combine the time series distribution characteristic data to establish a potential distribution model for each code value, and calculate the distribution probability of the feature value in the differentiated time series interval using the formula: ; Generate time series latent distribution encoding values; in, Represents the coded value The probability density function of Representative The risk weight of the coded value, For the The average outlier of the coded values, For the The standard deviation of the coded values, is the total number of coded values, and the exponential function is used to describe the distribution probability of the coded values.

[0034] formula: ; The benefit of the formula is that by introducing the risk weight parameter and distribution characteristic parameters , The collaborative calculation can reflect the probability characteristics of the potential distribution of the coding value in the time series signal, improve the accuracy of anomaly detection and the interpretability of the distribution characteristics of the time series signal; Detailed explanation of the formula and the process of formula calculation and derivation: set up , represents the risk weight of a certain code value, which is obtained through statistics of historical abnormal data; , is the average distribution characteristic of the code value; let , is the standard deviation of the coded value, indicating its distribution variability; let , is the current encoding value; substitute into the formula: ; ; ; ; The result shows that the distribution probability of the code value 55 is 0.02817, which is a low-probability distribution event. It is strongly correlated with the time series potential distribution code value result and can be further used for the detection and analysis of abnormal signal changes.

[0035] See also Figure 5 , the specific steps for obtaining the time series dynamic trend forecast value are: S401: Based on the data of the time series potential distribution coding value, the time series data of the pulse wave velocity is coded and inputted, the dynamic change rate at each time point is calculated, the change trend is analyzed according to the time series fluctuation, and the dynamic change rate data is generated; First, the code value of each time point is extracted and its time series is constructed. Then, the change rate of the code value in each time period is calculated, and the code value is segmented into small intervals to handle the dynamic time series characteristics. According to the code change rate in the time series, the dynamic change rate is calculated by comparing the difference between the code values ​​before and after, and the difference formula is used to calculate the dynamic change rate. Calculate the difference point by point and obtain the complete rate of change sequence. Further perform frequency statistics of positive and negative fluctuations on the rate of change sequence, and normalize it in combination with the frequency distribution to eliminate the dimension difference. The change rate data is standardized, and the standardized dynamic change rate series is used to evaluate the fluctuation characteristics of each interval in the time series and generate complete change rate data.

[0036] S402: Analyze the nonlinear offset characteristics of the dynamic change rate data, combine the time series fluctuation trend results, perform segmented evaluation on the nonlinear offset value, optimize the data trend by adjusting the offset distribution, and generate nonlinear trend analysis data; Firstly, the dynamic change rate series is segmented by fluctuation characteristics, and the time series is divided into multiple sub-intervals of equal length. Then, the mean and standard deviation in each interval are extracted as the initial description of the offset characteristics. Combined with the fluctuation amplitude characteristics, the characteristic data is fitted by a nonlinear regression model. The model adopts In the form of interval-by-interval optimization parameters The value of is used to fit the offset characteristics, and the offset amplitude and the change trend within the interval are further evaluated. The offset characteristics of each interval are weighted and summarized through the above calculation results to generate a comprehensive offset characteristic table containing a description of the nonlinear characteristics. The comprehensive characteristic table is used for the next trend analysis.

[0037] S403: combining nonlinear trend analysis data, integrating trend change values ​​of differentiated time series points, adjusting multi-layer trend distribution based on hierarchical modeling, processing the interaction between trend points, and generating merged trend change data; First, the trend change values ​​at different time points are integrated, and the trend of the time series is segmented into multiple levels. The trend fitting and weighted adjustment are performed on each level respectively. ,in is the level weight, It is a hierarchical function. By setting the hierarchical range and calculating the cumulative impact of each level, the overall trend change distribution is obtained. For the interactive impact between hierarchical points, the interactive formula is further used to calculate the correction term. , the interaction term is added to the trend adjustment to obtain the merged trend distribution containing the trend change correction value, which is further used as the main trend description of the dynamic change model.

[0038] S404: Based on the merged trend change data, calculate the correction parameter concentration fluctuation error using the formula: ; Generate time series dynamic trend forecast values; in, Representative time The trend forecast value of is the overall correction factor, For the The trend weight of the time point, For the The adjustment factor for the time point, is the total number of time points, and the exponential function describes the nonlinear impact of trend changes at time points on the overall trend.

[0039] formula: ; The benefit of the formula is that the introduction of exponential factors enables enhanced modeling of time trends for nonlinear changes, while the contribution of trend points is adjusted by weight parameters, thereby improving the dynamic adaptability and overall accuracy of the prediction.

[0040] Detailed explanation of the formula and the process of formula calculation and derivation: Representative time The trend forecast value of is the overall correction coefficient, which is derived by summarizing the dynamic characteristics of the comprehensive time series. Indicates The trend weight of a time point is calculated from the importance score of the trend point in the hierarchical trend analysis. For the The adjustment factor at a time point is evaluated based on the nonlinear offset characteristics. is the total number of time points, assuming , the values ​​are , obtained by dynamic data normalization , the adjustment factor is , the correction factor is .

[0041] Substitute the formula into the calculation item by item: ; ; ; calculate : ; Calculate each item step by step: ; Bring it into calculation: ; ; calculate : ; Calculate each item step by step: ; Bring it into calculation: ; ; calculate : ; Calculate each item step by step: ; Bring it into calculation: ; ; Summarizing the results: ; Result description: This result shows that the trend forecast value changes over time It shows nonlinear growth. The sensitivity of the model to dynamic trends is enhanced by introducing adjustment factors and exponential functions in the calculation. These predicted values ​​are further used to construct future trend models of time series to verify the accuracy of dynamic characteristic predictions.

[0042] See also Figure 6 The specific steps for obtaining the combined index value of cardiovascular risk trend are as follows: S501: Using the time series dynamic trend prediction value and the structural risk weight value matrix, cross-analyze the blood pressure waveform and heart rate variability parameters, extract the correlation of fluctuation data and evaluate key interaction points, and analyze the parameter interaction trend changes to generate cross-analysis data; The peak value of the blood pressure waveform and the interval change rate of the heart rate variability parameter are extracted as input data, and they are normalized respectively. Then, based on the correlation analysis, the interaction points of different parameters are determined. By counting the frequency of each interaction point in the overall data, the key interval change value is determined, and the interaction points with higher frequency are marked as high-interaction data. At the same time, the quadratic regression model is used to fit the trend change of the key interaction points. The regression equation is adopted ,in It is estimated by the least squares method, and the trend analysis of the data is carried out in combination with the fitting results to generate cross-analysis data.

[0043] S502: Based on the cross-analysis data, the dynamic change trend of key risk factors is extracted, and the characteristic fluctuation of each risk factor is evaluated in combination with the dynamic change curve. The dynamic data of key risk factors is generated by calculating the deviation rate and trend fitting value of key points; Combined with the dynamic change curve, the fluctuation value of each risk factor in the time dimension is used as input data, and the change rate of each factor is analyzed segment by segment. Calculate the rate of change of the time series, and then divide the fluctuation type of the factor according to the continuity and stability of the rate, and use the standard deviation formula The dynamic characteristics of the factors are quantified, and the relative importance of different risk factors in time series changes is evaluated. The quantified results are classified into three types: significant fluctuations, slight fluctuations, and no fluctuations. The overall trend is fitted through the classified fluctuation characteristics to generate dynamic data of key risk factors.

[0044] S503: Combine the dynamic data of key risk factors, call the waveform analysis parameters and integrate the risk trend, and at the same time, jointly evaluate the deviation correction model of the data, use the parameter distribution to adjust the weight value and optimize the trend distribution result, and generate the corrected evaluation model data; First, the dynamic changes of each risk factor are compared with the trend distribution parameters, and the quadratic deviation formula is calculated. Get the overall fluctuation range, and then use the normalization method to adjust the weight of each risk factor on the overall trend. The original values ​​are converted and a weighted model is constructed based on the converted values. The weight factor is calculated based on the mean and standard deviation of the dynamic trend. The weight correction is added to the parameters that deviate from the dynamic trend. The weight distribution of the overall risk trend is optimized through the corrected dynamic trend distribution to generate the corrected assessment model data.

[0045] S504: Calculate the comprehensive change trend of risk factors in the time series through the corrected evaluation model data, using the formula: ; Generate a combined index value for cardiovascular risk trend; in, Representative time The trend value of cardiovascular risk is the normalization coefficient, For the The weight of the risk factor, The peak risk time point is is the time diffusion parameter, is the total number of risk factors, and the exponential function describes the nonlinear impact of time changes in risk factors on the trend value.

[0046] formula: ; The benefit of the formula is that by introducing the weight of risk factors , time diffusion parameter And the normalization coefficient The time dynamic characteristics of multiple factors are comprehensively considered to enhance the adaptability and accuracy of the risk prediction model; Detailed explanation of the formula and the process of formula calculation and derivation: Setting the number of risk factors , whose weight is , , , diffusion parameter , , , peak time point , , , normalized coefficient ; About time point Perform the calculation: ; Itemized calculation: ; ; ; Summarizing the results: ; The results showed that the time point The combined index value of cardiovascular risk trend is 0.504, which belongs to the medium-to-high risk range of dynamic assessment, indicating that the risk value at time point 8 needs to be paid special attention to, and its calculation result is directly involved in the output and adjustment of the final trend indicator.

[0047] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A cardiovascular data monitoring method for cardiovascular medicine, characterized in that: The following steps are involved: S1: Based on the variable set of cardiovascular high-dimensional data, the characteristic parameters of blood pressure waveform and pulse wave velocity are sparsely regularized. The linear relationship is analyzed variable by variable, weak correlation edges are screened in descending order, redundant variables are eliminated and a dependency matrix is ​​established. The weight distribution is extracted from the relationship matrix to generate sparse dependency weight values. S2: Based on the sparse dependency weight value, the relationship between heart rate variability and arteriosclerosis index is distributedly fitted, the deviation amount and distribution parameters of multiple variables are updated, the strong correlation between variables is optimized, the relationship matrix is ​​reconstructed, and the structural risk weight value matrix is ​​generated; S3: Extracting features from the time series data of the electrocardiogram signal in segments based on the structural risk weight value matrix Change amount, encode the features and mark the abnormal intervals, analyze the time series changes based on the characteristics of the encoding value, and establish the time series potential distribution encoding value; S4: Based on the time series potential distribution coding value, the dynamic change rate of the implicit state is calculated using the coding input of the pulse wave velocity, the nonlinear offset characteristics are analyzed, the trend change values ​​are merged through hierarchical modeling, the fluctuation error in the parameter set is corrected, and the time series dynamic trend prediction value is obtained; S5: Based on the time series dynamic trend prediction value and the structural risk weight value matrix, cross-analysis of blood pressure waveform and heart rate variability parameters is performed to extract the dynamic change trend of key risk factors, and the joint evaluation index is calibrated to generate a joint index value of cardiovascular risk trend, implement individualized risk warning, optimization of intervention measures, and remote monitoring management.

2. The cardiovascular data monitoring method of cardiology department according to claim 1, characterized in that: The sparse dependency weight value includes weight distribution, sparse characteristics, and dependency matrix. The structural risk weight value matrix includes strong correlation between variables, deviation distribution parameters, and optimized relationship matrix. The time series potential distribution coding value includes characteristic change amount, abnormal interval annotation value, and coding value characteristic analysis result. The time series dynamic trend prediction value includes dynamic change rate, nonlinear offset characteristics, and trend change value. The cardiovascular risk trend joint indicator value includes dynamic change trend, key risk factors, and joint evaluation indicator correction results.

3. The cardiovascular data monitoring method of cardiology department according to claim 2, characterized in that: The specific steps for obtaining the sparse dependency weight value are as follows: S101: Based on the variable set of cardiovascular high-dimensional data, the initial characteristic parameters of blood pressure waveform and pulse wave velocity are extracted, the linear correlation between the variables and the waveform characteristics is analyzed one by one, the degree of dependence between the variables and the target characteristics is judged based on the correlation value, and the weakly correlated variables with correlation coefficients lower than the set threshold are screened to generate a weakly correlated variable set; S102: performing weighted calculation on multiple variables in the weakly associated variable set according to the distribution characteristics of their influence on the pulse wave velocity, calculating the redundancy value of each variable in the variable set, and eliminating variables with higher redundancy by comparing the redundancy value with a set threshold, thereby generating a variable set after eliminating redundancy; S103: In the variable set after removing redundancy, analyze the dependency relationship between variables, calculate the influence of variables on matrix weight distribution, construct a dependency matrix and extract weights, using the formula: ； Generate sparse dependency weight values; in, Representative variables and The dependency weight value of Representative variables and The dependency value of Representative variables and The distribution characteristic value of is the number of variables in the variable set, represents the variable being analyzed, As a mediating variable to measure the association, Another variable representing the dependency relationship.

4. The cardiovascular data monitoring method of cardiology department according to claim 3, characterized in that: The steps for obtaining the structural risk weight value matrix are specifically as follows: S201: Based on the sparse dependency weight value, analyzing the relationship between heart rate variability and arteriosclerosis index, extracting variable parameters associated with the two, calculating regression coefficients of multiple parameters in a multivariate relationship, adjusting parameter ranges using fitting errors, and generating a preliminary relationship distribution model; S202: Based on the preliminary relationship distribution model, update the deviation of multiple parameters in the model, call the updated parameters to calculate their influence on the multivariate distribution, optimize the distribution parameter range through difference analysis, establish updated distribution characteristics, and generate an updated relationship distribution model; S203: using the updated relationship distribution model, calculating the dependency strength between variables, adjusting the distribution of strongly associated variables according to the influence of the variables on the weights in the relationship matrix, reconstructing the dependency matrix, and generating an optimized relationship matrix; S204: Based on the optimized relationship matrix, extract the influence weights of the variables in the matrix on the risk distribution, using the formula: ； Calculate and generate the structural risk weight value matrix; in, Representative variables and The structural risk weight value of Representative variables With variables The dependency value of Representative variables The actual impact weight, Representative variables and The theoretical impact distribution weight of is the total number of variables, and the summation symbol indicates the cumulative risk value of all dependencies between the variables.

5. The cardiovascular data monitoring method of cardiology department according to claim 4, characterized in that: The steps for obtaining the temporal potential distribution coding value are specifically as follows: S301: Based on the structural risk weight value matrix, extract the characteristic variation within a time period from the time series data of the electrocardiogram signal, analyze the abnormal intervals of the signal characteristic variation within multiple time periods, call the extracted characteristic variation and mark the abnormal fluctuation range through segmented comparison to generate characteristic variation data; S302: digitally encode the characteristic variation data, calculate the corresponding relationship between the encoding value and the change of the electrocardiogram time series waveform, generate the distribution characteristic data of each feature segment in combination with the marked abnormal interval, mark the abnormal encoding value range in the corresponding time period, and generate the encoded abnormal interval data; S303: combining the encoded abnormal interval data, calling the distribution comparison result between the abnormal coding value and the normal coding value, evaluating the time series fluctuation characteristics and establishing the distribution characteristics of each segment of data, and generating time series distribution characteristic data; S304: Based on the time series distribution characteristic data, a potential distribution model for each code value is established, and the distribution probability of the feature value in the differentiated time series interval is calculated using the formula: ； Generate time series latent distribution encoding values; in, Represents the coded value The probability density function of Representative The risk weight of the coded value, For the The average outlier of the coded values, For the The standard deviation of the coded values, is the total number of coded values, the exponential function is used to describe the distribution probability of the coded values, and the denominator is the normalization term for the normalization of the distribution.

6. The cardiovascular data monitoring method of cardiology department according to claim 5, characterized in that: The steps for obtaining the time series dynamic trend prediction value are specifically as follows: S401: Based on the data of the time series potential distribution code value, the time series data of the pulse wave velocity is coded and inputted, the dynamic change rate at each time point is calculated, the change trend is analyzed according to the time series fluctuation, and the dynamic change rate data is generated; S402: Analyze the nonlinear offset characteristics of the dynamic change rate data, combine the time series fluctuation trend results, perform segmented evaluation on the nonlinear offset value, optimize the data trend by adjusting the offset distribution, and generate nonlinear trend analysis data; S403: Integrate the trend change values ​​of the differentiated time series points in combination with the nonlinear trend analysis data, adjust the multi-layer trend distribution based on hierarchical modeling, process the interaction between trend points, and generate merged trend change data; S404: Based on the merged trend change data, calculate the correction parameter centralized fluctuation error using the formula: ； Generate time series dynamic trend forecast values; in, Representative time The trend forecast value of is the overall correction factor, For the The trend weight of the time point, For the The adjustment factor for the time point, is the total number of time points, and the exponential function describes the nonlinear impact of trend changes at time points on the overall trend.

7. The cardiovascular data monitoring method of cardiology department according to claim 6, characterized in that: The steps for obtaining the cardiovascular risk trend joint index value are specifically as follows: S501: using the time series dynamic trend prediction value and the structural risk weight value matrix, cross-analyzing the blood pressure waveform and the heart rate variability parameters, extracting the correlation of the fluctuation data and evaluating the key interaction points, and analyzing the parameter interaction trend changes to generate cross-analysis data; S502: extracting the dynamic change trend of key risk factors based on the cross-analysis data, evaluating the characteristic fluctuation of each risk factor in combination with the dynamic change curve, and generating key risk factor dynamic data by calculating the deviation rate and trend fitting value of key points; S503: combining the dynamic data of the key risk factors, calling the waveform analysis parameters and integrating the risk trends, and combining the deviation correction model of the assessment data, adjusting the weight values ​​by using the parameter distribution and optimizing the trend distribution results, and generating the corrected assessment model data; S504: Calculate the comprehensive change trend of risk factors in the time series using the corrected assessment model data, using the formula: ； Generate a combined index value for cardiovascular risk trend; in, Representative time The trend value of cardiovascular risk is the normalization coefficient, For the The weight of the risk factor, The peak risk time point is is the time diffusion parameter, is the total number of risk factors, and the exponential function describes the nonlinear impact of time changes in risk factors on the trend value.

Images