Hypertensive cardiovascular risk hierarchical evaluation system based on intelligent decision support

Abstract

The invention discloses a hypertensive cardiovascular risk hierarchical assessment system based on intelligent decision support, and relates to the technical field of hypertensive cardiovascular risk assessment, and the system comprises a data collection module which is used for collecting health data of a patient measured through multi-brand wearable equipment in a family environment, and sending the health data to a database; the health data comprises blood pressure, heart rate, activity amount and sleep quality. According to the hypertensive cardiovascular risk hierarchical evaluation system based on intelligent decision support, the problem that data formats of wearable devices of multiple brands are not uniform can be solved, and the technical scheme of accurate hierarchical evaluation of the hypertensive cardiovascular risk is realized in combination with the intelligent decision support system, so that the health management level of the hypertensive is improved; the risk of cardiovascular diseases is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of cardiovascular risk assessment for hypertensive patients, and in particular to a cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support. Background Art

[0002] Hypertension is a common risk factor for cardiovascular disease and poses a serious threat to patients' cardiovascular health. With the development of science and technology, wearable devices have been widely used in the field of health monitoring, and patients can measure their blood pressure, heart rate and other health data at home. However, there are many brands of wearable devices on the market, and their data formats vary, which poses a great challenge to the unified processing and analysis of data. In addition, in a home environment, it is crucial to accurately assess the cardiovascular risk stratification of hypertensive patients due to the lack of guidance from professional medical staff. Traditional assessment methods often rely on patients to go to medical institutions for regular check-ups, but this method has certain limitations, such as limited detection frequency and inability to obtain real-time health data of patients in their daily state.

[0003] Existing technologies make it difficult to effectively integrate data from wearable devices of different brands, and are unable to fully utilize these data to achieve accurate stratified assessment of cardiovascular risk in patients with hypertension. Therefore, there is an urgent need to solve the problem of inconsistent data formats of multi-brand wearable devices and combine them with intelligent decision support systems to achieve technical solutions for accurate stratified assessment of cardiovascular risk in patients with hypertension, so as to improve the health management level of patients with hypertension and reduce the risk of cardiovascular disease. Summary of the invention

[0004] In order to solve the above technical problems, the present invention is implemented through the following technical solutions: a cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support, comprising: A data collection module, used to collect health data measured by patients in a home environment through multi-brand wearable devices, the health data including blood pressure, heart rate, activity level and sleep quality; The identification and parsing module is used to identify the source device brand and data format of the received health data, perform brand identification on the received health data based on the built-in brand database, match the corresponding data parsing template, standardize the health data in different formats into a unified system data format, and obtain standardized health data; The storage and processing module is used to receive the standardized health data processed by the identification and analysis module, and clean and preprocess the standardized health data, including outlier removal, missing value filling and data time series reconstruction, so as to obtain high-quality continuous patient data sets and obtain multi-dimensional data features; An intelligent assessment module, connected to the storage and processing module, is used to comprehensively assess the patient's current health status and cardiovascular risk level based on a predefined assessment model, wherein the assessment model is trained through a machine learning algorithm, combines the patient's personal basic information with real-time health data, outputs the patient's cardiovascular risk stratification results, obtains the assessment results, and generates health management recommendations based on the assessment results; The feedback reminder module is connected to the intelligent assessment module, generates feedback information based on the assessment results and health management suggestions, and feeds back the feedback information to the patient in a graphical form; at the same time, when it is detected that the patient's health data exceeds the safety threshold, it generates warning information and sends it to the patient and his family doctor for early warning, so as to carry out health intervention in time; Through fully automatic data collection and processing, combined with an intelligent decision support system, the problem of inconsistent data formats of multi-brand wearable devices in home scenarios is solved, and accurate stratified assessment of cardiovascular risks in patients with hypertension is achieved.

[0005] Preferably, the identification and analysis module includes: A brand identification unit is used to receive health data transmitted from a wearable device, analyze its data structure and feature code, identify the device brand of the data source based on a built-in brand database, and associate and match the device brand with a data parsing template to obtain a matching result; The data parsing unit is connected to the brand identification unit and is used to call the corresponding data parsing template based on the matching result of the device brand, extract the health data features in the health data, perform structured processing, and convert the extracted health data features into a system standard data format to obtain standardized health data.

[0006] Preferably, the identification and analysis module further includes: The exception handling unit is used to handle data missing or abnormal situations that occur during data analysis, and to perform data compensation and abnormal repair based on historical data and equipment characteristics to ensure that the quality of the output health data meets the evaluation requirements; The identification and analysis module can effectively respond to wearable devices of different brands and versions through multi-level data analysis and intelligent analysis, ensuring the accuracy and consistency of health data collection.

[0007] Preferably, the intelligent assessment module includes: A model training unit, used for training an evaluation model based on historical data of a large number of hypertensive patients, wherein the evaluation model includes linear regression, logistic regression, decision tree, random forest and neural network, and is used for capturing the relationship between the multidimensional data characteristics of the patient and the cardiovascular risk; The risk calculation unit is connected to the model training unit and is used to perform real-time calculations on the patient's current health data based on the trained assessment model, and output the patient's cardiovascular risk level in combination with the patient's basic personal information and environmental factors to obtain an assessment result; The intelligent assessment module achieves accurate prediction and assessment of patients' cardiovascular risks through model integration and personalized analysis.

[0008] Preferably, the basic personal information includes age, gender, weight, and medical history.

[0009] Preferably, the intelligent evaluation module further includes: The assessment unit is used to generate health management suggestions based on the patient's individual characteristics based on the assessment results of the risk calculation unit, so as to help the patient better manage his or her health status; the health management suggestions include diet suggestions, exercise suggestions, and drug management reminders; among which, individual characteristics include age, gender, and living habits.

[0010] Preferably, the feedback reminder module includes: The real-time feedback unit generates feedback information by graphically displaying the cardiovascular risk assessment results and combining them with the health management suggestions generated by the intelligent assessment module. The feedback information is sent to the patient in a graphical form. The feedback information includes the current blood pressure status, heart rate fluctuations, cardiovascular risk level and corresponding health management suggestions. The intelligent reminder unit is connected to the real-time feedback unit and is used to generate health reminders based on the results of cardiovascular risk assessment. When it is detected that the patient's health data exceeds the safety threshold, it automatically generates warning information and sends it to the patient, his family and family doctor. Among them, health reminders include reminders for taking medicines, regular physical examinations and lifestyle adjustment suggestions.

[0011] Preferably, the feedback reminder module further includes: The report generation unit is used to regularly summarize the patient's health status and cardiovascular risk changes over a period of time, generate the patient's health report, and send the patient's health report to the patient and family doctor respectively, so that the patient and doctor can track and adjust the health management effect; The feedback reminder module ensures that patients and their family doctors can obtain key health information in a timely manner and take appropriate intervention measures through multi-channel interaction.

[0012] Preferably, the storage processing module includes: The data cleaning unit is used to perform quality checks on the received standardized health data, remove outliers, fill in missing values, and smooth the data to ensure the continuity and reliability of the data; The data storage unit is used to store the cleaned health data in the patient's personal database, and the personal database is used to store the patient's long-term health data to support the dynamic update and evaluation of the model.

[0013] Preferably, the storage processing module further includes: The time series reconstruction unit is used to reconstruct and aggregate the patient's health data in time series. Through the data reconstruction technology based on the time window, the daily, weekly and monthly health data are summarized and the trend characteristics of the data are extracted to evaluate and track the patient's long-term health status; The storage and processing module provides high-quality health data input for the intelligent assessment module through comprehensive data cleaning and time series reconstruction, ensuring the credibility and accuracy of the assessment results.

[0014] The present invention provides a cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support, which has the following beneficial effects: This cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support has effectively solved the problem of inconsistent data formats of multi-brand wearable devices in home scenarios by developing universal data conversion middleware and identification and parsing modules; the module has built-in data format parsing templates for a variety of common wearable devices, which can automatically identify device brands or data features and match corresponding templates, convert data into a unified format, ensure that the system accurately receives and processes data, greatly improve the accuracy and consistency of data collection, and provide a reliable data foundation for subsequent accurate evaluation; it can solve the problem of inconsistent data formats of multi-brand wearable devices, and combine with intelligent decision support systems to realize technical solutions for accurate stratified assessment of cardiovascular risks in hypertensive patients, so as to improve the health management level of hypertensive patients and reduce the risk of cardiovascular disease. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flowchart of the cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support. DETAILED DESCRIPTION

[0016] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0017] like Figure 1As shown, the present invention provides a technical solution: a cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support, comprising: The data collection module is used to collect health data measured by patients through multi-brand wearable devices in a home environment, including blood pressure, heart rate, activity level and sleep quality; Specifically, the data collection module establishes a connection with wearable devices of various brands through Wi-Fi. The wearable devices collect the patient's health data at a frequency of once every two minutes and transmit the health data to the data collection module through Wi-Fi. These wearable devices are used by patients in a home environment to measure health data. The identification and parsing module is used to identify the source device brand and data format of the received health data, perform brand identification on the received health data based on the built-in brand database, match the corresponding data parsing template, standardize the health data in different formats into a unified system data format, and obtain standardized health data; The storage and processing module is used to receive the standardized health data processed by the identification and analysis module, and clean and preprocess the standardized health data, including outlier removal, missing value filling and data time series reconstruction, so as to obtain high-quality continuous patient data sets and obtain multi-dimensional data features; The intelligent assessment module is connected to the storage and processing module and is used to comprehensively assess the patient's current health status and cardiovascular risk level based on a predefined assessment model. The assessment model is trained through a machine learning algorithm and combines the patient's personal basic information with real-time health data to output the patient's cardiovascular risk stratification results, obtain the assessment results, and generate health management recommendations based on the assessment results; The feedback reminder module is connected to the intelligent assessment module, generates feedback information based on the assessment results and health management suggestions, and feeds back the feedback information to the patient in a graphical form. At the same time, when it is detected that the patient's health data exceeds the safety threshold, it generates early warning information and sends it to the patient and his family doctor for early warning so that timely health intervention can be carried out. Through fully automatic data collection and processing, combined with an intelligent decision support system, it solves the problem of inconsistent data formats of multi-brand wearable devices in home scenarios, and realizes accurate stratified assessment of cardiovascular risks in hypertensive patients.

[0018] The feedback reminder module provides patients with graphical feedback on assessment results and health management recommendations, allowing them to intuitively understand their own cardiovascular health status. At the same time, when it is detected that the patient's health data exceeds the safety threshold, early warning information can be generated in a timely manner and sent to the patient and his or her family doctor, realizing real-time monitoring and active intervention of the patient's health status, helping to prevent the disease from worsening and ensuring the patient's life safety.

[0019] The recognition and analysis modules include: A brand identification unit is used to receive health data transmitted from a wearable device, analyze its data structure and feature code, identify the device brand of the data source based on a built-in brand database, and associate and match the device brand with a data parsing template to obtain a matching result; The data parsing unit is connected to the brand identification unit and is used to call the corresponding data parsing template based on the matching result of the device brand, extract the health data features in the health data, perform structured processing, and convert the extracted health data features into a system standard data format to obtain standardized health data.

[0020] It should be further explained that, in the specific implementation process: the template is called based on the matching result: after receiving the associated matching result between the device brand and the data parsing template transmitted by the brand identification unit, the data parsing unit calls the corresponding data parsing template according to the matched device brand, and the data parsing template contains the parsing rules and extraction methods for the data format of a specific brand of wearable devices; wherein, the data parsing template is a set of rules pre-defined for the data format of a brand of wearable devices, which is used to accurately identify, extract and convert the health data features from the brand of devices to make them conform to the system standard data format, so as to realize the unified processing and analysis of the data of wearable devices of different brands, and provide an accurate and reliable data basis for the cardiovascular risk stratification assessment of hypertensive patients; Data feature extraction: Extract features from health data according to the rules of the called data parsing template: For health data containing blood pressure, heart rate, activity level, and sleep quality information, accurately identify and extract key feature values ​​from the received health data based on the rules of the data parsing template; if there is noise or interference information in the data, filter out irrelevant content based on the rules in the parsing template to obtain accurate health data features; Structural processing: When the extracted health data features contain scattered and unorganized data, the data parsing unit will perform structural processing on these features according to the pre-defined structural requirements of the system, including classifying blood pressure values ​​and heart rate values ​​into corresponding fields, and adding necessary identification information to each data feature, including measurement timestamp and data source device number, so that subsequent modules can clearly understand and process these data; Format conversion to standard format: Finally, the structured health data features are converted according to the standard data format specified by the system; among them, the system standard data format has unified requirements for data type, data length, and encoding method; including converting blood pressure values ​​into floating point numbers and expressing them in the unit mmHg specified by the system; through such format conversion, it is ensured that all data from wearable devices of different brands have a unified format when entering subsequent modules, and standardized health data is obtained, thereby realizing the system's accurate processing and analysis of data, and providing a reliable data basis for cardiovascular risk stratification assessment in hypertensive patients.

[0021] Among them, the data parsing template includes format parsing rules, feature extraction methods and data conversion specifications. Among them, the format parsing rules include data structure definition, field identification method and encoding parsing rules; the feature extraction method includes key feature positioning and data cleaning rules; the data conversion specifications include target format definition and unit conversion rules.

[0022] Through these elements, the data parsing template realizes the conversion process from the raw health data of a specific brand of wearable devices to standardized health data that can be processed by the system. It is a key part of achieving compatibility and effective utilization of multi-brand device data in the entire cardiovascular risk stratification assessment system for hypertensive patients.

[0023] The recognition and analysis module also includes: The exception handling unit is used to handle data missing or abnormal situations that occur during data analysis, and perform data compensation and exception repair based on historical data and equipment characteristics to ensure that the quality of the output health data meets the evaluation requirements.

[0024] It should be further explained that in the specific implementation process, the following situations should be considered when dealing with missing data: Missing data detection: First, the exception handling unit will conduct a comprehensive check on the health data generated during the data analysis process to determine the missing data. This is achieved by checking whether a specific field in the data set is null and whether there is a break in the data sequence. This includes the case where a heart rate value is missing at a certain point in time in a continuous heart rate data sequence. Compensation strategy based on historical data: Simple mean: For cases with few missing data points, the mean of the feature can be calculated based on the historical data of the same patient, and the missing values ​​can be filled with the calculated result; including: checking the heart rate data of the patient in the same period of the past week, calculating its average value to fill the current missing heart rate value; Time series interpolation method: For data with time series characteristics, an interpolation algorithm is used; including: if the blood pressure values ​​of two adjacent time points are known, but a time point in the middle is missing, use linear interpolation to estimate the missing value based on the known data points; Model-based prediction filling: Use machine learning models to build a data prediction model based on the patient's historical data and activity level; when missing data is encountered, other current related features are input into the model to predict the missing health data value; including: training a neural network model based on the patient's past blood pressure, heart rate, and activity level data; when blood pressure values ​​are missing, input the current heart rate and activity level features into the model to obtain the predicted blood pressure value for filling; Repair of abnormal data: Abnormal data identification: abnormal data is determined by comparing with the normal physiological range and judging based on the characteristics of the device; including: the systolic blood pressure of a normal adult is between 90-140 mmHg. If the systolic blood pressure is detected to be 200 mmHg, which is far beyond this range and does not conform to the patient's historical data trend, it is judged as abnormal data; at the same time, considering the characteristics of the device, inaccurate measurement results may be generated when the device is low in power. If it is found that the device power is at a low level and the data is abnormal when measuring data, it will also be marked as abnormal data; Repair strategies based on historical data and equipment characteristics: Data smoothing: For some outliers that may be caused by measurement errors, a data smoothing algorithm is used; including: weighted averaging of blood pressure data at several consecutive time points to reduce the impact of outliers on the overall data trend and make the data smoother and more reasonable.

[0025] Equipment calibration correction: When it is determined based on equipment characteristics that abnormal data may be caused by uncalibrated equipment, corrections are made by referring to equipment calibration information and historical calibration data; including: when it is known that temperature changes of the equipment in the past have affected the measurement accuracy, the measurement value is corrected by a specific calibration coefficient to adjust the abnormal blood pressure or heart rate data to a reasonable range; Correction based on historical data trends: Analyze the changing trends of the patient's historical data. If the current abnormal data deviates greatly from the trend, make reasonable corrections based on the trend. For example, the patient's blood pressure data over the past week showed a slow upward trend, but a certain measurement value suddenly increased significantly and did not conform to the trend. The abnormal data can be adjusted based on the trend prediction value to make it more consistent with the patient's actual health status. Data verification and rationality check: After completing data compensation and anomaly repair, the processed data is verified and rationally checked again; ensure that the filled or corrected data is within the physiological range and consistent with the patient's overall health status and historical data trends; if there are still unreasonable data points, further analysis of the reasons may be required, including re-evaluating the accuracy of the device data and considering the impact of the patient's recent lifestyle changes on the data. Manual intervention may be required when necessary, such as prompting the patient to remeasure and contacting medical staff for further diagnosis to ensure that the quality of the health data finally output meets the assessment requirements and provide reliable data support for the cardiovascular risk stratification assessment of hypertensive patients.

[0026] It should be further explained that, in the specific implementation process, data reception and preliminary analysis: the brand identification unit first receives the health data transmitted from the wearable device, which contains various measurement values ​​and data structures and feature codes that hide the device brand information; analyzes the received health data, including studying its data structure and feature codes, where the data structure includes the organization of the data, field definitions, and encoding rules; the feature code is a specific identifier embedded in the data by the device manufacturer, which is used to quickly identify the source of the device; Matching based on brand database: The brand recognition unit accesses its own built-in brand database, which stores relevant information of many common wearable device brands, including the data structure characteristics and feature code patterns of each brand of device; compares the structure and feature code of the received data with the records in the brand database; finds the best matching device brand record by comparing the features of the data with the feature information of each brand in the database one by one; Association matching and result generation: When the brand record that best matches the received data features is found, the brand identification unit will associate the device brand with the corresponding data parsing template and generate a matching result; the data parsing template is a pre-defined set of rules for parsing the data format of a specific brand of device; The generated matching results clearly indicate the device brand to which the received health data belongs and the corresponding data parsing template that should be used, so that the subsequent data parsing unit can accurately extract the health data features and perform format conversion; if no completely matching record is found in the brand database, the exception handling process will be triggered: further analyze the data features, prompt the user to check the device connection and data transmission problems, and try to use a common parsing strategy to ensure the continuity and accuracy of data processing.

[0027] The identification and analysis module can effectively respond to wearable devices of different brands and versions through multi-level data analysis and intelligent analysis, ensuring the accuracy and consistency of health data collection.

[0028] Smart assessment modules include: Model training unit, used to train and evaluate models based on historical data of a large number of hypertensive patients. Machine learning models include linear regression, logistic regression, decision tree, random forest and neural network, which are used to capture the relationship between multidimensional data characteristics of patients and cardiovascular risk; The risk calculation unit is connected to the model training unit and is used to perform real-time calculations on the patient's current health data based on the trained assessment model, and output the patient's cardiovascular risk level in combination with the patient's personal basic information and environmental factors to obtain an assessment result; wherein, the personal basic information includes age, gender, weight, and medical history; the intelligent assessment module achieves accurate prediction and assessment of the patient's cardiovascular risk through model integration and personalized analysis.

[0029] Based on the evaluation results, the system generates personalized health management recommendations for patients, including dietary recommendations, exercise recommendations, medication management reminders, etc., and takes into account the individual characteristics of the patients; this helps patients better understand their own health status, actively participate in self-health management, improve treatment compliance, and promote the development of a healthy lifestyle, thereby effectively controlling hypertension and reducing the risk of cardiovascular disease.

[0030] It should be further explained that in the specific implementation process, the construction process of the multi-parameter evaluation model is as follows: Data collection and preprocessing: A large amount of historical data of hypertensive patients is collected from multiple channels, including the patients' basic personal information and real-time health data measured by various methods; among them, basic personal information includes age, gender, weight, medical history, and real-time health data includes blood pressure, heart rate, activity level, and sleep quality; the collected data is strictly preprocessed, including data cleaning, outlier processing, and missing value filling to ensure the quality and accuracy of the data; multidimensional data features related to cardiovascular risk are extracted from the preprocessed historical data, including conversion, combination and screening of the original data; for example, the mean and variance of blood pressure and the fluctuation range of heart rate over a period of time are calculated, and the daily activity intensity is calculated based on the activity data; the extracted features are normalized or standardized to make different features comparable, so as to avoid affecting the model training effect due to excessive differences in the feature value range; for example, blood pressure values ​​and heart rate values ​​are mapped to between [0-1]; Model training, integration and evaluation: First, the model training unit is used to train linear regression, logistic regression, decision tree, random forest and neural network based on the historical data of a large number of hypertensive patients. During the training process, the model parameters are adjusted by optimizing their respective loss functions to achieve a better fitting effect on the training set.

[0031] After training, use an independent validation set to evaluate each model and record the performance indicators of each model on the validation set, including accuracy, recall, F1 value, and mean square error. These performance indicators will serve as an important basis for the weight allocation of each model in the subsequent integrated model. Weight determination: Determine the weight of each model in the integrated model based on its performance on the validation set: Use the weighted average method, and the weight calculation is based on the model's performance indicators; weights can be assigned based on the model's accuracy, and the higher the accuracy, the greater the weight assigned to the model; assuming that after evaluation, the highest accuracy of the neural network model on the validation set is 0.85, the accuracy of the logistic regression model is 0.78, the accuracy of the decision tree model is 0.75, the accuracy of the random forest model is 0.72, and the accuracy of the linear regression model is 0.70, then these accuracies can be attributed The weights are obtained by normalization. The specific method of calculating the weights is as follows: first, the sum of the accuracy of all models is calculated as 0.85+0.78+0.75+0.72+0.70=3.8; then the weight of the neural network model is 0.85 / 3.8≈0.224, the weight of the logistic regression model is 0.78 / 3.8≈0.205, the weight of the decision tree model is 0.75 / 3.8≈0.197, the weight of the random forest model is 0.72 / 3.8≈0.189, and the weight of the linear regression model is 0.70 / 3.8≈0.184.

[0032] Integrated prediction: When predicting new patient data, the patient's data is input into the five trained models at the same time; each model will output its own prediction results. For classification problems: for example, judging the cardiovascular risk level, belonging to high, medium, and low categories, the output is the category prediction probability; for regression problems: for example, predicting the probability value of cardiovascular disease, the output is a specific numerical prediction.

[0033] Then, the prediction results of each model are weighted and summed according to the previously determined weights. For example, for classification problems, the probability of each model predicting a certain category is multiplied by the corresponding weight and added together to obtain the final prediction probability of the integrated model for that category; for regression problems, the prediction values ​​of each model are multiplied by the weight and added together to obtain the final prediction value of the integrated model; Finally, based on the prediction results of the integrated model, such as the final risk level probability or predicted value, the patient's cardiovascular risk is assessed, thereby achieving the goal of improving the accuracy and reliability of the assessment through model integration, and providing more accurate decision support for the cardiovascular risk stratification assessment of hypertensive patients.

[0034] The Smart Assessment Module also includes: The assessment unit is used to generate health management suggestions based on the patient's individual characteristics based on the assessment results of the risk calculation unit, so as to help the patient better manage his or her health status; the health management suggestions include diet suggestions, exercise suggestions, and drug management reminders; among which, individual characteristics include age, gender, and living habits.

[0035] The intelligent assessment module is based on a pre-defined assessment model, comprehensively considering the patient's personal basic information and real-time health data, and using machine learning algorithm training models to capture the nonlinear relationship between multidimensional data characteristics and cardiovascular risk, thereby achieving accurate assessment of the patient's cardiovascular risk level; compared with traditional assessment methods, it can more comprehensively and accurately reflect the patient's actual cardiovascular risk status, help to detect potential high-risk patients early, and take timely intervention measures.

[0036] The feedback reminder module includes: The real-time feedback unit generates feedback information by graphically displaying the cardiovascular risk assessment results and combining them with the health management suggestions generated by the intelligent assessment module. The feedback information is sent to the patient in a graphical form. The feedback information includes the current blood pressure status, heart rate fluctuations, cardiovascular risk level and corresponding health management suggestions. The intelligent reminder unit is connected to the real-time feedback unit and is used to generate health reminders based on the results of cardiovascular risk assessment. When it is detected that the patient's health data exceeds the safety threshold, it automatically generates warning information and sends it to the patient, his family and family doctor. Among them, health reminders include reminders for taking medicines, regular physical examinations and lifestyle adjustment suggestions.

[0037] The feedback reminder module also includes: The report generation unit is used to regularly summarize the patient's health status and cardiovascular risk changes over a period of time, generate the patient's health report, and send the patient's health report to the patient and family doctor respectively, so that the patient and doctor can track and adjust the health management effect; The feedback reminder module ensures that patients and their family doctors can obtain key health information in a timely manner and take appropriate intervention measures through multi-channel interaction.

[0038] The storage processing module includes: The data cleaning unit is used to perform quality checks on the received standardized health data, remove outliers, fill in missing values, and smooth the data to ensure the continuity and reliability of the data; The data storage unit is used to store the cleaned health data in the patient's personal database. The personal database is used to store the patient's long-term health data to support the dynamic update and evaluation of the model.

[0039] The storage processing module also includes: The time series reconstruction unit is used to reconstruct and aggregate the patient's health data in time series. Through the data reconstruction technology based on the time window, the daily, weekly and monthly health data are summarized and the trend characteristics of the data are extracted to evaluate and track the patient's long-term health status.

[0040] It should be further explained that in the specific implementation process, the process of aggregating health data and extracting trend features is carried out through the data reconstruction technology based on the time window: Time window definition and data division: According to system requirements and data characteristics, time windows of different time scales are defined, including daily, weekly and monthly. For the daily time window, all health data collected in one day are regarded as one data unit, including multiple measurements of blood pressure, heart rate, activity level and sleep quality. The weekly time window covers all daily data units in a week, and the monthly time window is analogous. Continuous health data is divided according to these time windows to ensure that the data in each time window is relatively independent and complete, providing a basis for subsequent aggregation and analysis. For example, for blood pressure data recorded by the hour, 24 hours of data are classified as data in one daily time window. Data aggregation and aggregation operations: In each time window, various types of health data are aggregated and calculated; for numerical data, the average, sum, maximum, and minimum values ​​can be calculated; for example, in the daily time window, the average blood pressure, maximum and minimum values ​​of the heart rate, and the sum of the amount of activity are calculated for the day; among them, numerical data include blood pressure, heart rate, and amount of activity; for categorical data, the frequency of occurrence of each category can be counted; for example, the percentage of days with good sleep quality in a week can be counted; through these aggregation operations, a large amount of original health data is compressed into a set of representative statistical feature data in each time window, reducing the amount of data while retaining key information; among them, categorical data includes: sleep quality is divided into good, average, and poor; Trend feature extraction method: Simple trend analysis: Calculate the rate of change of statistical features between adjacent time windows to observe the upward or downward trend of the data; for example, calculate the difference between the average blood pressure this week and the average blood pressure last week, and divide it by the average blood pressure last week to get the blood pressure change rate, so as to determine whether the blood pressure is rising, falling or relatively stable during this period; for activity data, calculate the month-on-month growth rate of the total weekly activity to understand the trend of changes in the patient's activity level; Moving average method: Use moving average technology to further smooth data trends and reduce the impact of short-term fluctuations on trend judgment; for example, calculate the moving average of the average blood pressure per month over the past three months. As time goes by, the moving average of each month will reflect a longer-term and more stable blood pressure trend. For activity data, you can also calculate moving averages of different time spans, such as the moving average of the past two weeks or the past month, to better observe the long-term trend of activity. Polynomial fitting and trend modeling: Use polynomial fitting methods to model statistical feature data within a time window to extract more complex trend features; including quadratic polynomial fitting of the average heart rate data per month over the past year to obtain the coefficients of the fitting curve, which can characterize the trend change pattern of heart rate data in the long term; in this way, nonlinear trends hidden in the data can be discovered, providing richer information for accurately assessing the patient's long-term health status; Seasonal analysis: If there are seasonal changes in health data, including the situation that patients have relatively high blood pressure in winter and lower blood pressure in summer, the data can be decomposed into trend items, seasonal items and random items through time series decomposition methods; for monthly or quarterly data, analyze the change patterns of seasonal items and understand the typical characteristics of patients' health data in different seasons. This will help to more comprehensively evaluate the relationship between patients' cardiovascular health status and seasonal factors, and take corresponding health management measures in advance; for example, if it is found that patients' blood pressure is significantly higher seasonally in winter every year, blood pressure monitoring and management can be strengthened in winter.

[0041] Feature integration and application: Integrate the trend features extracted from different time windows; different time windows include daily, weekly, and monthly, forming a feature vector containing multi-time scale trend information; for example, a feature vector may contain the trend change rate of average blood pressure this month, the moving average of activity in the past three months, and the polynomial fitting coefficient of heart rate in the past year.

[0042] These integrated trend characteristics will serve as important inputs for subsequent assessment and tracking of patients' long-term health status, providing more comprehensive and time-dimensional information support for the intelligent assessment module in the cardiovascular risk stratification assessment system for hypertensive patients, helping to more accurately predict patients' cardiovascular risk levels, formulate personalized health management recommendations, and promptly detect potential health problems and intervene.

[0043] It should be further explained that, in the specific implementation process, the storage and processing module provides high-quality health data input for the intelligent assessment module through comprehensive data cleaning and time series reconstruction, ensuring the credibility and accuracy of the assessment results; The storage and processing module comprehensively cleans, preprocesses and reconstructs the patient's health data in time series. It aggregates daily, weekly and monthly health data through time window-based data reconstruction technology, extracts trend features, and provides high-quality data input for the intelligent assessment module, thereby achieving effective tracking and analysis of the patient's long-term health status. Doctors and patients can use this to understand the development trend of the disease, adjust health management strategies, and improve the effectiveness and pertinence of disease management.

[0044] Automatic data collection and processing in home scenarios reduces the need for patients to frequently visit medical institutions for examinations, alleviating the burden on medical institutions. At the same time, it enables doctors to remotely understand the patient's condition in real time, rationally allocate medical resources, and manage and intervene in patients more efficiently, thereby optimizing the medical service model and improving the overall level of medical services.

[0045] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support, characterized in that: include: A data collection module, used to collect health data measured by patients in a home environment through multi-brand wearable devices, the health data including blood pressure, heart rate, activity level and sleep quality; The identification and parsing module is used to identify the source device brand and data format of the received health data, perform brand identification on the received health data based on the built-in brand database, match the corresponding data parsing template, standardize the health data in different formats into a unified system data format, and obtain standardized health data; The storage and processing module is used to receive the standardized health data processed by the identification and analysis module, and clean and preprocess the standardized health data, including outlier removal, missing value filling and data time series reconstruction to obtain multi-dimensional data features; An intelligent assessment module, connected to the storage and processing module, is used to comprehensively assess the patient's current health status and cardiovascular risk level based on a predefined assessment model, wherein the assessment model is trained through a machine learning algorithm, combines the patient's personal basic information with real-time health data, outputs the patient's cardiovascular risk stratification results, obtains the assessment results, and generates health management recommendations based on the assessment results; The feedback reminder module is connected to the intelligent assessment module, generates feedback information based on the assessment results and health management suggestions, and feeds back the feedback information to the patient in a graphical form; at the same time, when it is detected that the patient's health data exceeds the safety threshold, it generates warning information and sends it to the patient and his family doctor for warning.

2. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 1, characterized in that: The identification and analysis module includes: A brand identification unit is used to receive health data transmitted from a wearable device, analyze its data structure and feature code, identify the device brand of the data source based on a built-in brand database, and associate and match the device brand with a data parsing template to obtain a matching result; The data parsing unit is connected to the brand identification unit and is used to call the corresponding data parsing template based on the matching result of the device brand, extract the health data features in the health data, perform structured processing, and convert the extracted health data features into a system standard data format to obtain standardized health data.

3. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 2, characterized in that: The identification and analysis module also includes: The exception handling unit is used to handle data missing or abnormal situations that occur during data analysis, and perform data compensation and exception repair based on historical data and equipment characteristics to ensure that the quality of the output health data meets the evaluation requirements.

4. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 3 is characterized by: The intelligent evaluation module comprises: A model training unit, used for training an evaluation model based on historical data of hypertensive patients, wherein the evaluation model includes linear regression, logistic regression, decision tree, random forest and neural network; The risk calculation unit is connected to the model training unit and is used to perform real-time calculations on the patient's current health data based on the trained assessment model, and output the patient's cardiovascular risk level in combination with the patient's basic personal information and environmental factors to obtain an assessment result.

5. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 4 is characterized in that: The basic personal information includes age, gender, weight, and medical history.

6. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 5, characterized in that: The intelligent assessment module also includes: The evaluation unit is used to generate health management suggestions according to the individual characteristics of the patient based on the evaluation results of the risk calculation unit; the health management suggestions include diet suggestions, exercise suggestions, and drug management reminders; among which, individual characteristics include age, gender, and living habits.

7. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 6, characterized in that: The feedback reminder module includes: The real-time feedback unit generates feedback information by graphically displaying the cardiovascular risk assessment results and combining them with the health management suggestions generated by the intelligent assessment module. The feedback information is sent to the patient in a graphical form. The feedback information includes the current blood pressure status, heart rate fluctuations, cardiovascular risk level and corresponding health management suggestions. The intelligent reminder unit is connected to the real-time feedback unit and is used to generate health reminders based on the results of cardiovascular risk assessment. When it is detected that the patient's health data exceeds the safety threshold, a warning message is generated and sent to the patient, his family and family doctor. Among them, health reminders include reminders for taking medicines, reminders for regular physical examinations and suggestions for lifestyle adjustments.

8. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 7, characterized in that: The feedback reminder module also includes: The report generation unit is used to regularly summarize the patient's health status and changes in cardiovascular risks over a period of time, generate the patient's health report, and send the patient's health report to the patient and family doctor respectively, so that the patient and doctor can track and adjust the health management effect.

9. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 8, characterized in that: The storage processing module includes: The data cleaning unit is used to perform quality checks on the received standardized health data, remove outliers, fill in missing values, and smooth the data; The data storage unit is used to store the cleaned health data in the patient's personal database, and the personal database is used to store the patient's long-term health data.

10. The cardiovascular risk stratification assessment system for hypertensive patients based on intelligent decision support according to claim 9, characterized in that: The storage processing module also includes: The time series reconstruction unit is used to reconstruct and aggregate the patient's health data in time series. Through the data reconstruction technology based on the time window, the daily, weekly and monthly health data are summarized and the trend characteristics of the data are extracted to evaluate and track the patient's long-term health status.

Images