An ePWV correction evaluation method and system based on a hypertensive population

Abstract

The present application relates to the technical field of blood pressure risk assessment, in particular to an ePWV correction evaluation method and system based on a hypertensive population, which comprises the following steps: collecting static data and dynamic data of a patient, generating corresponding blood pressure data sequences according to the time sequence of the dynamic data corresponding to each static data; configuring a threshold adjustment area for the blood pressure data sequence, and determining the blood pressure variability of the blood pressure sequence through the threshold adjustment area; taking the blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics and blood pressure circadian rhythm as dynamic characteristic data; inputting the static data and dynamic characteristic data into a pre-trained ePWV correction model to obtain a correction result; comparing the correction result with a risk threshold, and marking the patient corresponding to the correction result exceeding the risk threshold as having a vascular abnormality risk, which can improve the accuracy of the blood pressure variability coefficient and thus improve the accuracy of the vascular abnormality risk assessment.

Description

Technical Field

[0001] This invention relates to the field of vascular abnormality risk assessment technology, specifically to an ePWV correction assessment method and system based on hypertensive individuals. Background Technology

[0002] Currently, ambulatory blood pressure monitoring (ABPM) is the primary technique for assessing blood pressure variability. Existing technologies typically calculate blood pressure variability by performing 24-hour continuous ambulatory blood pressure monitoring, collecting blood pressure data at multiple time points, and calculating statistical indicators such as standard deviation and coefficient of variation based on the collected blood pressure data sequences to characterize the degree of blood pressure variability. However, current technologies implicitly consider fluctuations in all blood pressure data sequences as pathological blood pressure variability. In fact, numerous medical studies have shown that patient behavior (such as changes in posture, exercise, diet, and emotional fluctuations) and environmental factors (such as temperature changes and noise exposure) significantly affect instantaneous blood pressure measurements. Current technologies cannot distinguish between behavior / environment-driven physiological blood pressure fluctuations and pathological blood pressure variability, leading to inaccurate blood pressure variability assessments.

[0003] To address these issues, we propose a modified ePWV assessment method and system for hypertensive individuals. Summary of the Invention

[0004] The purpose of this invention is to provide a modified ePWV assessment method and system for hypertensive patients to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method and system for corrective assessment of ePWV in hypertensive patients, the method comprising the following steps: Collect static and dynamic data from patients, and generate a corresponding blood pressure data sequence from the dynamic data according to the collection time series for each static data point; A threshold adjustment region is configured for the blood pressure data sequence. The threshold adjustment region includes a behavioral correlation map, a data chain, and an adjustment chain. The blood pressure variability of the blood pressure sequence is determined through the threshold adjustment region. Blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics, and blood pressure diurnal rhythm are used as dynamic feature data. Static and dynamic feature data are input into a pre-trained ePWV correction model to obtain correction results. The correction results are compared with a risk threshold, and patients whose correction results exceed the risk threshold are marked as having a risk of vascular abnormalities.

[0006] Preferably, the step of configuring a threshold adjustment area for the blood pressure data sequence includes: setting multiple data points corresponding to the blood pressure data value at each collection time point, and connecting the multiple data points in the order of the collection time points to obtain a data chain; setting an adjustment chain for the data chain, wherein the adjustment chain includes multiple information points and adjustment points; obtaining patient behavior information corresponding to multiple historical blood pressure collection time points, constructing a behavior association map based on the patient behavior information at multiple collection time points, establishing the association relationship between the behavior association map and the data chain, and using the behavior association map, the data chain, and the association relationship as the threshold adjustment area.

[0007] Preferably, the steps of acquiring patient behavior information corresponding to multiple historical blood pressure collection time points, constructing a behavior association map based on the patient behavior information from multiple collection time points, and establishing the association between the behavior association map and the data chain include: acquiring the types of multiple historical patient behavior information and the corresponding standard values ​​of blood pressure data, wherein the patient behavior information includes action feature information and environmental feature information, setting multiple information nodes corresponding to the types of patient behavior information, acquiring the same feature information among multiple patient behavior information, and associating the information nodes corresponding to multiple patient behavior information based on the same feature information to generate a behavior association map.

[0008] Preferably, the step of setting a regulation chain for the data chain, wherein the regulation chain includes multiple information points and regulation points, includes: obtaining the blood pressure data standard values ​​corresponding to each patient behavior information, setting information points corresponding to multiple blood pressure data standard values, sorting the multiple information points in order according to the size of the stored multiple blood pressure data standard values ​​to generate a regulation chain, setting regulation points for the regulation chain, and establishing the connection relationship between the regulation points and multiple information points.

[0009] Preferably, the step of determining the blood pressure variability of a blood pressure sequence through a threshold adjustment region includes: setting monitoring points for the patient; acquiring blood pressure data values ​​and corresponding patient behavior information in real time based on the monitoring points; storing the blood pressure data values ​​sequentially on data points in the data chain according to the order of acquisition time points; establishing communication relationships between the monitoring points and multiple information nodes; acquiring the current blood pressure data value corresponding to which the blood pressure variability needs to be calculated, as well as the current patient behavior information corresponding to the current blood pressure data value; comparing the patient behavior information corresponding to the current blood pressure data value with the patient behavior information corresponding to multiple information nodes according to the communication relationships; obtaining the blood pressure data standard value corresponding to the patient behavior information with the highest similarity as the target blood pressure data standard value corresponding to the current blood pressure data value; adjusting the adjustment point on the information point to the corresponding target information point according to the target blood pressure data standard value; and determining the blood pressure variability based on the target blood pressure data standard value and the current blood pressure data value.

[0010] Preferably, the step of obtaining the blood pressure data standard value corresponding to the patient behavior information with the highest similarity as the target blood pressure data standard value corresponding to the current blood pressure data value includes: obtaining the current patient behavior information corresponding to the current collection time point, and the previous patient behavior information corresponding to the previous collection time point; comparing the current patient behavior information with the previous patient behavior information to obtain the current identical feature information; marking the corresponding information nodes in the behavior association graph according to the previous patient behavior information; determining the patient behavior information corresponding to other information nodes associated with the marked information nodes through the behavior association graph, as the preferred patient behavior information; comparing the current patient behavior information with the preferred patient behavior information, determining the historical patient behavior information consistent with the current patient behavior information from multiple historical patient behavior information according to the behavior association graph, and using the blood pressure data standard value corresponding to the historical patient behavior information as the target blood pressure data standard value.

[0011] Preferably, the step of inputting static data and dynamic feature data into a pre-trained ePWV correction model to obtain correction results; comparing the correction results with a risk threshold, and marking patients whose correction results exceed the risk threshold as having a risk of vascular abnormalities includes: Static data and dynamic feature data are concatenated to form a joint input feature vector, which is then input into a pre-trained ePWV correction evaluation model to output the corrected estimated pulse wave velocity. A preset risk threshold library is obtained, which stores at least one vascular abnormality risk threshold. The corrected estimated pulse wave conduction velocity is numerically compared with the vascular abnormality risk threshold. When the corrected estimated pulse wave velocity exceeds the vascular abnormality risk threshold, the target object is marked as having a vascular abnormality risk.

[0012] A modified ePWV assessment system for hypertensive populations, applied to any of the above-described modified ePWV assessment methods for hypertensive populations, comprising: The data acquisition module is used to collect static and dynamic data from patients. For each static data point, the dynamic data is used to generate a corresponding blood pressure data sequence according to the collection time series. The feature extraction module is used to configure a threshold adjustment region for the blood pressure data sequence. The threshold adjustment region includes a behavioral correlation map, a data chain, and an adjustment chain. The blood pressure variability of the blood pressure sequence is determined through the threshold adjustment region. Blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics, and blood pressure diurnal rhythm are used as dynamic feature data. The risk assessment module is used to input static and dynamic feature data into a pre-trained ePWV correction model to obtain correction results; the correction results are compared with risk thresholds, and patients whose correction results exceed the risk thresholds are marked as having vascular abnormality risk.

[0013] Compared with the prior art, the beneficial effects of the present invention are: By constructing a behavioral association graph, the correlation between patient behavioral information and standard blood pressure data values ​​is established. Based on the current patient behavioral information, the corresponding target blood pressure standard value is dynamically matched, thereby effectively correcting behavioral / environmental interference in blood pressure variability calculation. The standard blood pressure data value can be determined based on the patient behavioral information at the time point of blood pressure data collection, ensuring that the obtained blood pressure variability better matches the blood pressure data values ​​collected at the corresponding time point, thus improving the accuracy of blood pressure variability calculation and consequently, the accuracy of vascular risk assessment. When the collected current behavioral information is not completely consistent with historical behavioral information, the behavioral association graph can quickly locate other associated information nodes through the marked information nodes, narrowing the priority matching range to the set of associated nodes. This allows for the rapid identification of the historical behavioral information closest to the current behavioral information and the corresponding blood pressure standard value, improving the efficiency of vascular risk assessment. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of 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.

[0015] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a system structure block diagram of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] For examples, please refer to Figures 1 to 2 This invention provides a method and system technical solution for ePWV correction assessment based on hypertensive patients: A method for ePWV correction assessment based on hypertensive patients includes the following steps: S1: Collect the patient's static and dynamic data, and generate a corresponding blood pressure data sequence for each static data point by generating the dynamic data according to the collection time series. Specifically, obtain the target object's static first dataset, including: age, gender, height, weight, whether they have diabetes, and whether they have chronic kidney disease. Obtain the target object's dynamic second dataset, namely, 24-hour ambulatory blood pressure monitoring sequence data (including timestamps, systolic blood pressure, diastolic blood pressure, and heart rate). S2: Configure a threshold adjustment region for the blood pressure data sequence. The threshold adjustment region includes a behavioral correlation map, a data chain, and a adjustment chain. The blood pressure variability of the blood pressure sequence is determined through the threshold adjustment region. Blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics, and blood pressure diurnal rhythm are used as dynamic feature data. Specifically, after denoising the dynamic second dataset, the following key indicators were calculated: Mean Arterial Pressure (MBP): 24-hour, daytime, and nighttime mean arterial pressure were calculated. Blood Pressure Variability (BPV): 24-hour systolic blood pressure coefficient of variation (SCV = SDsbp / Meansbp) and diastolic blood pressure coefficient of variation (DCV) were calculated. Heart Rate Fluctuation Characteristics: 24-hour mean heart rate and its standard deviation were calculated. Blood pressure diurnal rhythm: Calculate the rate of decline in blood pressure at night and label it as dipping, non-dipping, or reverse dipping; Blood pressure variability (BPV) refers to the calculation of the 24-hour systolic blood pressure coefficient of variation (SCV = SDsbp / Meansbp) and diastolic blood pressure coefficient of variation (DCV). Blood pressure abnormality is determined by comparing blood pressure data values ​​with standard blood pressure data values. Blood pressure variability is calculated based on blood pressure abnormality at multiple time points. Traditional BPV is calculated directly based on the original blood pressure values ​​(incorporating behavioral / environmental interference). This application calculates based on a "behavioral-corrected abnormality sequence" (removing behavioral / environmental interference). By increasing the accuracy of calculating blood pressure abnormality, the accuracy of blood pressure variability is improved, thereby improving the accuracy of vascular abnormality risk assessment.

[0018] The steps for configuring a threshold adjustment region for a blood pressure data sequence include: setting multiple data points corresponding to the blood pressure data value at each collection time point, connecting the multiple data points in the order of the collection time points to obtain a data chain; setting an adjustment chain for the data chain, wherein the adjustment chain includes multiple information points and adjustment points; obtaining patient behavior information corresponding to multiple historical blood pressure collection time points, constructing a behavior association map based on the patient behavior information at multiple collection time points, establishing the association relationship between the behavior association map and the data chain, and using the behavior association map, data chain, and association relationship as the threshold adjustment region; The steps of acquiring patient behavior information corresponding to multiple historical blood pressure collection time points, constructing a behavior association graph based on the patient behavior information from multiple collection time points, and establishing the association between the behavior association graph and the data chain include: acquiring the types of multiple historical patient behavior information and the corresponding standard values ​​of blood pressure data, wherein the patient behavior information includes action feature information and environmental feature information, setting multiple information nodes corresponding to the types of patient behavior information, acquiring the common feature information among multiple patient behavior information, and associating the information nodes corresponding to multiple patient behavior information based on the common feature information to generate a behavior association graph.

[0019] The steps of setting a regulation chain for the data chain, wherein the regulation chain includes multiple information points and regulation points, include: obtaining the blood pressure data standard values ​​corresponding to each patient behavior information, setting information points corresponding to multiple blood pressure data standard values, sorting the multiple information points in order according to the size of the stored multiple blood pressure data standard values ​​to generate a regulation chain, setting regulation points for the regulation chain, and establishing the connection relationship between the regulation points and multiple information points. It should be noted that the data chain is a sequence of blood pressure data values ​​connected in chronological order of acquisition time. Each data point corresponds to a blood pressure measurement value at a specific time point, essentially acting as a time-series database. Each data point stores the blood pressure data values ​​at each acquisition time point, and the data chain serves as the storage medium for the raw blood pressure data. The regulation chain is a dynamic parameter chain composed of information points and regulation points. It stores and locates the standard blood pressure values ​​corresponding to different behaviors, acting as a parameter server. It is an addressable parameter storage array that can adaptively select the corresponding standard blood pressure data values ​​at different time points. Information points are fixed-position nodes on the regulation chain. Each information point stores the standard blood pressure data value corresponding to a type of patient behavior information. Each information point stores the blood pressure data standard value for a specific patient type. The behavioral information corresponds to the standard blood pressure data value (such as the standard blood pressure value in a sedentary state, the standard blood pressure value after a meal, etc.); the adjustment point is a movable pointer / cursor on the adjustment chain, used to point to the currently active information point, representing the standard blood pressure value corresponding to the current behavior. It is like a virtual machine, which can move between multiple information points, which is equivalent to changing the communication relationship between the adjustment point and multiple information points, such as disconnection or connection. When the environment or behavior changes, the adjustment point automatically moves to the matching information point. The adjustment point is a movable pointer / cursor on the adjustment chain, pointing to the currently active information point. The position of the adjustment point determines the current reference standard value used to calculate blood pressure variability. The adjustment point is like a lightweight state machine or context environment handle. It does not store the data itself, but holds a reference to the currently valid parameter (information point). When patient behavior changes, the adjustment point "migrates" to a new information point, similar to the dynamic migration of a virtual machine between physical machines, but what is migrated is the "context pointer" rather than the entire runtime environment. Information nodes are nodes in the behavior association graph, representing a type of patient behavior information (action features or environmental features), equivalent to a graph database entity capable of storing multiple historical patient behavior information, their corresponding types, and the standard blood pressure data values ​​for that type. Monitoring points are the collection points that collect blood pressure data and patient behavior information in real time; they are the collection terminals of a wearable blood pressure device, such as a dynamic blood pressure monitoring device. Communication relationships are the logical connections between monitoring points and information nodes, used to match real-time behavior with historical behavior graphs. The behavior association graph is a graph structure where nodes are information nodes (representing a type of behavior), and edges indicate that two types of behavior share common feature information (such as both having the "sitting" or "post-meal" feature). This graph is used to quickly retrieve the behavior type most similar to the current behavior from historical behaviors. The adjustment point is a movable logical pointer unit whose position on the adjustment chain is dynamically bound to the current patient behavior type. The adjustment point holds a reference to the currently valid information point and automatically migrates along the adjustment chain to the target information point when the behavior changes.At the system architecture level, adjustment points can be instantiated as lightweight state machine contexts or parameter cursor registers. Their migration process is similar to context switching between different physical nodes in a virtual machine, but the migration granularity is parameter pointers rather than a complete runtime environment. For changed behavioral information detected by monitoring points, the adjustment chain is moved to the corresponding data point. This is equivalent to reactivating the current patient behavior information stored on the adjustment point and comparing it with various types of patient behavior information in the behavior association graph when patient behavior information changes. This avoids immediately comparing patient behavior information with historical patient behavior information based on changes in the collection time point; analysis and comparison are only performed when changes occur, reducing the use of computing resources and improving computational efficiency. The adjustment point and monitoring point are connected, and the current patient behavior information collected by the monitoring point is promptly transmitted to the adjustment point. When a monitoring point detects a change in the patient's behavior, it transmits this information to the control point. Simultaneously, a connection is established between the control point and the data point containing the current blood pressure data collected at the corresponding time point. This connection acts as a communication node between the data point and the control point. One communication node moves along the control chain, and the communication node and the control chain are bound together. The other communication node is the data point. The control point can drive the control chain to move across multiple data points, effectively establishing a communication connection between the control point and its corresponding data points. This reduces redundant time-series feature calculations and improves system processing efficiency.

[0020] The steps for determining blood pressure variability of a blood pressure sequence through threshold adjustment include: setting monitoring points for patients; acquiring blood pressure data values ​​and corresponding patient behavior information in real time based on the monitoring points; storing the blood pressure data values ​​sequentially on data points in the data chain according to the order of acquisition time; establishing communication relationships between monitoring points and multiple information nodes; acquiring the current blood pressure data value corresponding to which blood pressure variability needs to be calculated, as well as the current patient behavior information corresponding to the current blood pressure data value; comparing the patient behavior information corresponding to the current blood pressure data value with the patient behavior information corresponding to multiple information nodes according to the communication relationships; obtaining the blood pressure data standard value corresponding to the patient behavior information with the highest similarity as the target blood pressure data standard value corresponding to the current blood pressure data value; adjusting the adjustment point on the information point to the corresponding target information point according to the target blood pressure data standard value; and determining the blood pressure variability based on the target blood pressure data standard value and the current blood pressure data value. The steps for obtaining the blood pressure data standard value corresponding to the patient behavior information with the highest similarity as the target blood pressure data standard value corresponding to the current blood pressure data value include: obtaining the current patient behavior information corresponding to the current collection time point, and the previous patient behavior information corresponding to the previous collection time point; comparing the current patient behavior information with the previous patient behavior information to obtain the current identical feature information; marking the corresponding information nodes in the behavior association graph according to the previous patient behavior information; determining the patient behavior information corresponding to other information nodes associated with the marked information nodes through the behavior association graph, as the preferred patient behavior information; comparing the current patient behavior information with the preferred patient behavior information, determining the historical patient behavior information consistent with the current patient behavior information from multiple historical patient behavior information according to the behavior association graph, and using the blood pressure data standard value corresponding to the historical patient behavior information as the target blood pressure data standard value.

[0021] Specifically, when the monitoring points collect data at different times, the control chain moves across multiple data points. This is equivalent to binding the control chain to the corresponding data collection time points of the monitoring points. The standard blood pressure values ​​on the control chain differ depending on the time point. The standard blood pressure value is determined by the location of the control point. The standard blood pressure value corresponding to the control point is used as the standard blood pressure value for the corresponding data point. The information point where the control point is located is determined based on the patient behavior information collected from the blood pressure data value at that data point. The information point on the control chain containing the standard blood pressure value of patient behavior information consistent with the current data point's patient behavior information is used as the target information point for the control point. This is based on the current blood pressure data... Blood pressure variability is determined by comparing the measured blood pressure values ​​with the standard values ​​at the target information points. Since the standard thresholds for different blood pressure measurements may differ depending on the environment or behavior, different standard thresholds are set for the blood pressure data collected at corresponding time points based on the patient's current behavior and environment. This makes the obtained blood pressure variability more accurate, thereby improving the accuracy of arteriosclerosis risk assessment. The standard value for blood pressure data can be determined based on the patient's behavioral information at the time point of blood pressure data collection, so that the obtained blood pressure variability can better match the blood pressure data values ​​collected at the corresponding time point, thereby improving the accuracy of blood pressure variability calculation and thus improving the accuracy of vascular risk assessment.

[0022] The system acquires current patient behavior information at the current data collection time point and previous patient behavior information at the previous data collection time point. It compares the current patient behavior information with the previous patient behavior information to obtain information with the same characteristics. Based on the previous patient behavior information, it marks corresponding information nodes in a behavior association graph. The behavior association graph then identifies other information nodes associated with the marked information nodes as preferred patient behavior information. The system prioritizes comparing the current patient behavior information with the preferred patient behavior information. This allows for the rapid identification of historical patient behavior information consistent with the current patient behavior information from multiple historical patient behavior data points based on the behavior association graph. This leads to the rapid determination of the blood pressure data standard value corresponding to the current patient behavior information, thereby improving the efficiency of blood pressure variability calculation and vascular risk assessment. Furthermore, through the communication relationship between information nodes and adjustment points, the adjustment point is moved to the information point corresponding to the blood pressure data standard value to improve the accuracy of subsequent calculations of patient blood pressure variability.

[0023] S3: Input static data and dynamic feature data into the pre-trained ePWV correction model to obtain the correction result; compare the correction result with the risk threshold, and mark patients whose correction results exceed the risk threshold as having a risk of vascular abnormalities; The steps of inputting static and dynamic feature data into a pre-trained ePWV correction model to obtain correction results, and comparing the correction results with risk thresholds to mark patients whose correction results exceed the risk thresholds as having a risk of vascular abnormalities include: concatenating static and dynamic feature data to form a joint input feature vector, inputting it into a pre-trained ePWV correction assessment model to output the corrected estimated pulse wave velocity; obtaining a preset risk threshold library, which stores at least one vascular abnormality risk threshold; comparing the corrected estimated pulse wave velocity with the vascular abnormality risk threshold; and marking the target object as having a risk of vascular abnormalities when the corrected estimated pulse wave velocity exceeds the vascular abnormality risk threshold.

[0024] Specifically, the risk threshold library includes multiple graded vascular abnormality risk thresholds: First risk threshold: 8 m / s ≤ modified ePWV < 10 m / s, corresponding to "moderate risk of vascular dysfunction"; Second risk threshold: 10 m / s ≤ modified ePWV < 12 m / s, corresponding to "high risk of vascular abnormality"; Third risk threshold: modified ePWV ≥ 12 m / s, corresponding to "extremely high risk of vascular abnormality"; and outputs the corresponding vascular abnormality risk level label based on the threshold range in which the modified ePWV falls.

[0025] Specifically, the training of the ePWV correction model involves: acquiring a training dataset, which includes several training samples. Each training sample contains static clinical parameters, dynamic hemodynamic feature parameters, and corresponding measured carotid-femoral pulse wave velocity (cfPWV) values. The static clinical parameters include at least age, height, and diabetes prevalence markers. The dynamic hemodynamic feature parameters include at least mean arterial pressure (MBP) and systolic blood pressure coefficient of variation (SCV). The training dataset is divided into a training set and a validation set. The measured cfPWV values ​​are used as supervision labels, and the static clinical parameters and dynamic hemodynamic feature parameters are used as input features. A pre-defined machine learning algorithm or... A regression analysis algorithm is used to train the model, resulting in an initial ePWV correction evaluation model. The validation set is used to evaluate the performance of the trained initial ePWV correction evaluation model, calculating the mean absolute error (MAE) or correlation coefficient between the model's predicted values ​​and the measured cfPWV values. When the mean absolute error is less than a preset threshold or the correlation coefficient is greater than a preset threshold, the current model is determined as the final ePWV correction evaluation model. The training dataset is derived from clinical data collection from a specific population, namely the Chinese hypertensive population. The measured cfPWV value of each training sample is obtained using the gold standard method. 70% of the training dataset is divided into a training set and 30% into a validation set.

[0026] An improved nonlinear regression equation algorithm was used for model training, specifically including: constructing an initial regression equation, the general form of which is: ePWV_pred=f(Age,MBP)+α·Height+β·SCV+γ·Diabetes+δ, where: f(Age,MBP) is a nonlinear basic function based on age and mean arterial pressure; α is the height correction coefficient; β is the systolic blood pressure coefficient of variation (SCV) correction coefficient; γ is the diabetes correction term; δ is the intercept term; using measured cfPWV as the target variable, and age, mean arterial pressure, height, systolic blood pressure coefficient of variation, and diabetes markers from the training set as independent variables, multivariate nonlinear regression analysis was used to solve for the coefficients α, β, γ, δ and the parameters of the basic function f in the regression equation; the solved coefficients and parameters were stored in the ePWV correction evaluation model, and the correction coefficients obtained by multivariate regression analysis were: height correction coefficient α=-0.05; systolic blood pressure coefficient of variation correction coefficient β=0.15; diabetes correction term γ=0.8; Performance evaluation includes at least one of the following validation metrics: mean absolute error (MAE); root mean square error (RMSE); Pearson correlation coefficient (r) between predicted and measured cfPWV; and the percentage of samples with an absolute difference ≤ 1 m / s. The model is considered successfully trained when the mean absolute error on the validation set is ≤ 1.0 m / s and the correlation coefficient r ≥ 0.75. The trained ePWV modified evaluation model is compared with an existing European population ePWV model on the same validation set. The mean absolute error of the two models is calculated. The superiority of this model is confirmed when its mean absolute error is lower than that of the European model.

[0027] A modified ePWV assessment system for hypertensive populations, applied to any of the above-described modified ePWV assessment methods for hypertensive populations, comprising: The data acquisition module is used to collect static and dynamic data from patients. For each static data point, the dynamic data is used to generate a corresponding blood pressure data sequence according to the collection time series. The feature extraction module is used to configure a threshold adjustment region for the blood pressure data sequence. The threshold adjustment region includes a behavioral correlation map, a data chain, and an adjustment chain. The blood pressure variability of the blood pressure sequence is determined through the threshold adjustment region. Blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics, and blood pressure diurnal rhythm are used as dynamic feature data. The risk assessment module is used to input static and dynamic feature data into a pre-trained ePWV correction model to obtain correction results; the correction results are compared with risk thresholds, and patients whose correction results exceed the risk thresholds are marked as having vascular abnormality risk.

[0028] By constructing a behavioral association graph, the correlation between patient behavioral information and blood pressure data standard values ​​is established. Based on the current patient behavioral information, the corresponding target blood pressure standard value is dynamically matched, thereby effectively correcting behavioral / environmental interference in blood pressure variability calculation. The system can determine the corresponding blood pressure data standard value based on the patient behavioral information at the time point of blood pressure data collection, ensuring that the obtained blood pressure variability better matches the blood pressure data value collected at the corresponding time point, thus improving the accuracy of blood pressure variability calculation and consequently improving the accuracy of vascular risk assessment. When the collected current behavioral information is not completely consistent with historical behavioral information, the behavioral association graph can be used to quickly locate other associated information nodes through the marked information nodes, narrowing the priority matching range to the set of associated nodes. This allows for the rapid identification of the historical behavioral information closest to the current behavioral information and its corresponding blood pressure standard value, improving the efficiency of vascular risk assessment. A regulation chain structure containing multiple information points and regulation points is designed. Information points are sorted according to the size of the blood pressure standard value, and regulation points can be quickly located and migrated on the regulation chain based on the target blood pressure standard value corresponding to the current patient behavioral information. By binding the regulation chain to the data chain and establishing communication relationships between regulation points and information points, the system achieves dynamic and adaptive adjustment of the blood pressure abnormality judgment threshold at different data collection time points and under different behavioral states. Binding the regulation chain to the data collection time point realizes the spatiotemporal correlation of behavioral information, data collection time point, and standard blood pressure data values. The system can automatically adjust the evaluation benchmark of corresponding data points based on changes in the patient's behavioral state at different time points, achieving truly individualized and time-series-based assessment of blood pressure variability. Through behavioral correlation mapping and dynamic threshold matching, the assessment of blood pressure variability no longer relies on expensive measurement equipment (such as continuous blood pressure monitors or arteriosclerosis detection equipment); high-precision assessment can be achieved with only routine ambulatory blood pressure monitoring data and patient behavior / environmental annotation information, significantly reducing assessment costs.

[0029] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0030] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A modified ePWV assessment method for hypertensive individuals, characterized in that, Includes the following steps: Collect static and dynamic data from patients, and generate a corresponding blood pressure data sequence from the dynamic data according to the collection time series for each static data point; A threshold adjustment region is configured for the blood pressure data sequence. The threshold adjustment region includes a behavioral correlation map, a data chain, and an adjustment chain. The blood pressure variability of the blood pressure sequence is determined through the threshold adjustment region. Blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics, and blood pressure diurnal rhythm are used as dynamic feature data. Static and dynamic feature data are input into a pre-trained ePWV correction model to obtain correction results. The correction results are compared with a risk threshold, and patients whose correction results exceed the risk threshold are marked as having a risk of vascular abnormalities.

2. The ePWV correction assessment method based on hypertensive populations according to claim 1, characterized in that: The step of configuring a threshold adjustment region for the blood pressure data sequence includes: Multiple data points are set for the blood pressure data value at each collection time point, and the multiple data points are connected in the order of collection time points to form a data chain; Set a control chain for the data link, where the control chain includes multiple information points and control points; Obtain patient behavior information corresponding to multiple historical blood pressure collection time points, construct a behavior association map based on the patient behavior information from multiple collection time points, establish the association relationship between the behavior association map and the data chain, and use the behavior association map, data chain, and association relationship as a threshold adjustment area.

3. The ePWV correction assessment method based on hypertensive populations according to claim 2, characterized in that: The steps for obtaining patient behavior information corresponding to multiple historical blood pressure collection time points, constructing a behavior association graph based on the patient behavior information from multiple collection time points, and establishing the association between the behavior association graph and the data chain include: The system acquires multiple types of historical patient behavior information and corresponding standard values ​​of blood pressure data. The patient behavior information includes action characteristic information and environmental characteristic information. Multiple information nodes are set according to the type of patient behavior information; the same feature information is obtained among multiple patient behavior information, and the information nodes corresponding to multiple patient behavior information are associated based on the same feature information to generate a behavior association map.

4. The ePWV correction assessment method based on hypertensive populations according to claim 3, characterized in that: Setting up a control chain for the data link, where the control chain includes multiple information points and control points, involves the following steps: Obtain the standard blood pressure data values ​​corresponding to each patient's behavioral information; Information points are set for multiple blood pressure data standard values, and the multiple information points are sorted in order of the magnitude of the stored multiple blood pressure data standard values ​​to generate a regulation chain; Set adjustment points for the adjustment chain and establish connection relationships between the adjustment points and multiple information points.

5. The ePWV correction assessment method based on a hypertensive population according to claim 4, characterized in that: The steps for determining blood pressure variability in a blood pressure sequence through a threshold adjustment region include: Monitoring points are set for patients, and blood pressure data values ​​and corresponding patient behavior information are obtained in real time based on the monitoring points. The blood pressure data values ​​are stored sequentially on the data points of the data chain according to the order of the collection time. Establish communication relationships between monitoring points and multiple information nodes; obtain the current blood pressure data value corresponding to the blood pressure variability to be calculated, as well as the current patient behavior information corresponding to the current blood pressure data value; and compare the patient behavior information corresponding to the current blood pressure data value with the patient behavior information corresponding to the multiple information nodes according to the communication relationships. Obtain the blood pressure data standard value corresponding to the patient behavior information with the highest similarity as the target blood pressure data standard value corresponding to the current blood pressure data value; The adjustment point is adjusted to the corresponding target information point based on the target blood pressure data standard value, and the blood pressure variability is determined based on the target blood pressure data standard value and the current blood pressure data value.

6. The ePWV correction assessment method based on a hypertensive population according to claim 5, characterized in that: The steps to obtain the blood pressure data standard value corresponding to the patient behavior information with the highest similarity as the target blood pressure data standard value corresponding to the current blood pressure data value include: Obtain current patient behavior information corresponding to the current collection time point, and previous patient behavior information corresponding to the previous collection time point; The current patient behavior information is compared with the previous patient behavior information to obtain the current identical feature information; the corresponding information nodes are marked in the behavior association graph according to the previous patient behavior information. The patient behavior information corresponding to other information nodes associated with the marked information nodes is determined by the behavior association graph and used as the preferred patient behavior information. The current patient behavior information is compared with the preferred patient behavior information. Based on the behavior association map, the historical patient behavior information that is consistent with the current patient behavior information is determined from multiple historical patient behavior information. The blood pressure data standard value corresponding to the historical patient behavior information is used as the target blood pressure data standard value.

7. The ePWV correction assessment method based on a hypertensive population according to claim 1, characterized in that: The static data and dynamic feature data are input into the pre-trained ePWV correction model to obtain the correction result; The steps of comparing the corrected results with a risk threshold and marking patients whose corrected results exceed the risk threshold as having a risk of vascular abnormalities include: Static data and dynamic feature data are concatenated to form a joint input feature vector, which is then input into a pre-trained ePWV correction evaluation model to output the corrected estimated pulse wave velocity. A preset risk threshold library is obtained, which stores at least one vascular abnormality risk threshold. The corrected estimated pulse wave conduction velocity is numerically compared with the vascular abnormality risk threshold. When the corrected estimated pulse wave velocity exceeds the vascular abnormality risk threshold, the target object is marked as having a vascular abnormality risk.

8. An ePWV correction assessment system for hypertensive populations, applied to the ePWV correction assessment method for hypertensive populations as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect static and dynamic data from patients. For each static data point, the dynamic data is used to generate a corresponding blood pressure data sequence according to the collection time series. The feature extraction module is used to configure a threshold adjustment region for the blood pressure data sequence. The threshold adjustment region includes a behavioral correlation map, a data chain, and an adjustment chain. The blood pressure variability of the blood pressure sequence is determined through the threshold adjustment region. Blood pressure variability, mean arterial pressure, heart rate fluctuation characteristics, and blood pressure diurnal rhythm are used as dynamic feature data. The risk assessment module is used to input static and dynamic feature data into a pre-trained ePWV correction model to obtain correction results; the correction results are compared with risk thresholds, and patients whose correction results exceed the risk thresholds are marked as having vascular abnormality risk.

Images