Health management data fusion method and system based on multiple intelligent wearable devices

Abstract

The invention relates to the technical field of health information management, and discloses a health management data fusion method and system based on multiple intelligent wearable devices, and the method comprises the steps: obtaining the management information of the intelligent wearable devices of a user, and constructing a health information collection network of the user; health management data is acquired based on an acquisition network, a health assessment AI model is utilized to perform preliminary assessment to form a health information feature matrix, use time feature analysis equipment use modes of all equipment are extracted, health trend analysis is performed through the health information feature matrix, and health trend fitting feature distribution is obtained; performing prediction analysis based on trend distribution, generating a prediction information feature matrix, performing information verification processing, feeding back and correcting a prediction result, forming health trend verification feature distribution, and finally integrating all health data and feature distribution to generate a health management map. The problem that in the prior art, the data analysis utilization rate of various intelligent wearable devices is low is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of health information management, and particularly to a health management data fusion method and system based on multiple intelligent wearable devices. Background Art

[0002] Currently, there are various types of intelligent wearable devices on the market, such as smart bracelets, smart watches, sports trackers, smart glasses, etc. These devices not only vary in type, but also the health parameters they monitor are different, including but not limited to exercise data, heart rate, blood sugar, sleep quality, blood oxygen, etc. Each device has differences in function and usage method, resulting in great differences in the formats, accuracies, and frequencies of the health data they generate. These data are diverse in type and large in quantity, making it difficult to effectively extract effective information from them to evaluate the user's health status and make long-term predictions. At the same time, health data is usually time-series data, and the user's health status changes over time. Existing data analysis lacks consideration of how to dynamically evaluate the health status in the time dimension, resulting in inaccurate evaluation results. Summary of the Invention

[0003] The purpose of the present invention is to provide a health management data fusion method and system based on multiple intelligent wearable devices, aiming to solve the problem of low data analysis and utilization rate of multiple intelligent wearable devices in the prior art.

[0004] The present invention is implemented as follows. In the first aspect, the present invention provides a health management data fusion method based on multiple intelligent wearable devices, including: Obtaining the management information of the user's intelligent wearable devices, and performing a relevance analysis on the health management data collection method for the user according to the intelligent wearable device management information to obtain the user's health information collection network; Obtaining the user's health management data based on the user's health information collection network, and performing a preliminary evaluation and information conversion on the health management data according to a pre-trained health assessment AI model to obtain the user's health information feature matrix; Extracting the usage time features and summarizing the patterns of each intelligent wearable device according to the health information feature matrix to obtain the device usage patterns of the user for each intelligent wearable device; Performing a trend analysis of the health status at the continuous time level on the health information feature matrix according to the device usage pattern to obtain the health trend fitting feature distribution of the user; Performing a prediction analysis on the health information feature matrix based on the health trend fitting feature distribution to obtain the prediction information feature matrix; Performing information verification processing on the health information feature matrix generated for future time according to the predicted information feature matrix, so as to perform feedback correction on the health trend fitting feature distribution according to the verification result, and obtain a health trend verification feature distribution; Performing multi-dimensional information integration on the health management data, the health information feature matrix, the health trend fitting feature distribution, and the health trend verification feature distribution to obtain a health management map for providing feedback on the user's health data.

[0005] In a second aspect, the present invention provides a health management data fusion system based on multiple intelligent wearable devices, which is used to implement the health management data fusion method based on multiple intelligent wearable devices described in any one of the first aspects.

[0006] The present invention provides a health management data fusion method based on multiple intelligent wearable devices, which has the following beneficial effects: The present invention obtains the intelligent wearable device management information of the user to construct the health information collection network of the user, obtains health management data based on the collection network, and uses the health assessment AI model for preliminary assessment to form a health information feature matrix, extracts the usage time features of each device to analyze the device usage pattern, performs health trend analysis through the health information feature matrix to obtain a health trend fitting feature distribution, performs prediction analysis based on the trend distribution to generate a predicted information feature matrix, and performs information verification processing, feeds back and corrects the prediction result to form a health trend verification feature distribution, and finally integrates all health data and feature distributions to generate a health management map, solving the problem of low data analysis and utilization rate of multiple intelligent wearable devices in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 is a step schematic diagram of a health management data fusion method based on multiple intelligent wearable devices provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0008] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0009] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0010] Referring to Figure 1 as shown, a preferred embodiment is provided by the present invention.

[0011] In a first aspect, the present invention provides a health management data fusion method based on multiple intelligent wearable devices, including: S1: Obtain the management information of the user's smart wearable device, and perform a correlation analysis on the health management data collection method for the user based on the management information of the smart wearable device to obtain the user's health information collection network; S2: Obtain the user's health management data based on the user's health information collection network, and perform a preliminary evaluation and information conversion on the health management data according to a pre-trained health assessment AI model to obtain the user's health information feature matrix; S3: Extract the usage time features and summarize the patterns for each smart wearable device according to the health information feature matrix to obtain the device usage patterns of the user for each smart wearable device; S4: Perform a trend analysis of the health status at the continuous time level on the health information feature matrix according to the device usage pattern to obtain the health trend fitting feature distribution of the user; S5: Perform a predictive analysis on the health information feature matrix based on the health trend fitting feature distribution to obtain a predictive information feature matrix; S6: Perform information verification processing on the health information feature matrix generated at future times according to the predictive information feature matrix, and perform feedback correction on the health trend fitting feature distribution according to the verification result to obtain a health trend verification feature distribution; S7: Perform multi-dimensional information integration on the health management data, the health information feature matrix, the health trend fitting feature distribution, and the health trend verification feature distribution to obtain a health management map for providing feedback on the user's health data.

[0012] Specifically, in step S1 of the embodiment provided by the present invention, the system needs to extract basic management information from the user's smart wearable device, including basic information such as device type, model, brand, usage frequency, usage duration, etc.; collect the connection methods (such as Bluetooth, Wi-Fi, NFC, etc.) between devices, as well as the data synchronization methods, times, and frequencies. These information helps to understand the accuracy and reliability of device data collection.

[0013] More specifically, obtain various types of sensors built into the device (such as heart rate monitors, gait sensors, accelerometers, etc.) and the types of health data they collect (such as steps, heart rate, sleep quality, blood oxygen concentration, etc.), and analyze the user's personalized configuration of the device, such as health goal setting, reminder mode, monitoring time period, etc. These are helpful for understanding the usage habits and data collection patterns of the device. Through this process, the system can comprehensively understand the configuration, usage method, and data types collected by the user's smart wearable device, laying a foundation for subsequent data analysis and health assessment.

[0014] More specifically, analyze the data interaction of different smart wearable devices, study the data sharing and correlation among devices. For example, a smartwatch may provide data on the user's steps and heart rate, while smart glasses may provide sleep quality data. Analyzing the correlation between them can reveal which device data is complementary and which is redundant. By analyzing the usage patterns of devices (such as usage frequency, duration, time period, etc.), determine which devices contribute the most to health data collection. Combining the context information of device usage, further establish a relationship model between each device and the health status.

[0015] More specifically, fuse the data from different devices to form a comprehensive health information data stream. Optimize the quality and accuracy of the data through algorithms (such as weighted algorithms, data cleaning algorithms, etc.). Through the above analysis, establish a health information collection network, clarify the role of each device in the network (such as data source, data center or processing node). Each device in the network represents a part of the health management system and is responsible for collecting or monitoring a certain type of health data. This analysis process helps to discover the correlation and interaction between different smart wearable devices, making the user's health data collection more systematic, optimized, and able to cover the user's health status more comprehensively. On this basis, the system can dynamically adjust the data collection strategy according to the contribution degree, accuracy, and reliability of different devices to ensure the comprehensiveness and accuracy of health data.

[0016] More specifically, based on the analysis results, optimize the node layout of the health information collection network. For example, if the health data collection effect of a certain device is not good, its role in the health management network can be adjusted or other devices can be used instead. Optimize the data collection frequency of each device according to the importance of the device and the type of data collected. For example, heart rate monitoring may require real-time collection, while step data can be processed by hourly or daily statistics. Dynamically adjust the device configuration in the health information collection network according to the user's health changes (such as changes in health goals or fluctuations in disease status) to ensure the real-time and accuracy of data collection and health assessment. The optimization of the health information collection network can ensure that each device is in the best working state, avoid redundant data collection or insufficient data collection, thereby improving the efficiency and accuracy of the entire health management system. The dynamic adjustment strategy can achieve personalized configuration for different users to adapt to different health needs and enhance the flexibility of the health management system.

[0017] More specifically, after completing the above analysis, the system will obtain a comprehensive and optimized health information collection network. This network will show the interrelationships among various devices, sensors, and collection methods, and clarify the role of each device in health data collection. Finally, the health information collection network can not only accurately collect various types of health data, but also form an efficient data circulation system through the complementary cooperation of multiple devices, providing accurate basic data support for subsequent health assessment, prediction, and feedback correction. This health information collection network not only improves the comprehensiveness, real-time nature, and accuracy of health data collection, but also provides a reliable data source for subsequent health trend analysis, assessment model, and health management atlas generation.

[0018] Specifically, in step S2 of the embodiment provided by the present invention, relevant health management data is collected from each device in the health information collection network, including multi-dimensional data such as heart rate, blood pressure, number of steps, sleep quality, exercise amount, body temperature, etc. Through the communication interfaces between devices, the collected data of each device is integrated into a centralized data platform in real time or regularly, and the obtained health management data is cleaned and preprocessed. This includes removing missing values, handling outliers, and data standardization (such as unifying the units of data from different devices) to ensure the quality and usability of the data. The data collected by the user's different devices is summarized into a standardized format for subsequent health assessment and processing. Data integration and cleaning can ensure the consistency of health data from different devices in terms of format, quality, and integrity, providing an accurate and reliable data basis for subsequent analysis. Through data standardization, the comparability of data from different devices under the same health management framework is ensured, avoiding errors caused by device differences.

[0019] More specifically, according to different types of health data and the user's health goals, a suitable pre-trained AI model is selected. For example, a cardiovascular health assessment model, a sleep quality assessment model, an exercise amount and weight management model, etc. The cleaned and standardized health data is input into the health assessment AI model. This data may include the user's daily number of steps, heart rate fluctuations, sleep duration, etc. The pre-trained AI model conducts a preliminary assessment of the input health data. The AI model may include a regression model, a classification model based on machine learning, or a neural network based on deep learning, etc., for assessing the user's health status, such as assessing whether there is a risk of heart disease, whether in a state of high blood pressure, etc.

[0020] More specifically, the AI model will transform the user's health data to generate a health assessment output, including the user's health risks, health indices, or other assessment results. The health assessment AI model can automatically conduct a preliminary assessment based on the user's health data, reducing manual intervention and improving the accuracy and efficiency of the assessment process. Through the automated processing of AI, the health data can be professionally evaluated, which helps to quickly identify the user's health risks (such as hypertension, diabetes, etc.) and provides support for subsequent health management decisions.

[0021] More specifically, based on the assessment results of the health assessment AI model, the user's health characteristic information is extracted. For example, cardiovascular health characteristics, exercise ability characteristics, sleep quality characteristics, etc. These characteristics usually include a set of numerical values or categories, representing different dimensions of the user's health status. The extracted health characteristic data is aggregated into a health information characteristic matrix. Each row in the matrix represents a time point (e.g., daily or weekly data), and each column represents a health characteristic (such as heart rate, number of steps, amount of exercise, etc.). This matrix synthesizes health data from different dimensions to form a comprehensive description of the user's health status.

[0022] More specifically, according to specific health management needs, the characteristic matrix can be further expanded. For example, adding the user's personalized needs (such as weight loss, muscle mass gain, etc.) or extending the user's health trends based on time series data. The process of generating the health information characteristic matrix can transform complex health data into a standardized matrix form, enabling various types of health data to be uniformly processed, analyzed, and compared. The health information characteristic matrix provides a multi-dimensional and structured health data model, facilitating subsequent trend analysis, prediction, health management decisions, etc. The matrix form can effectively aggregate information from different devices and health dimensions, helping analysts or AI systems to understand and evaluate the user's health status.

[0023] More specifically, based on the health information characteristic matrix, the system can conduct time series analysis on the dynamic changes in the user's health to identify health trends, such as trends in weight loss, heart rate changes, etc. Through further analysis of the characteristic matrix, personalized health advice can be provided to the user. For example, based on the cardiovascular health assessment results, a suitable exercise plan and diet regulation are recommended. Through machine learning or deep learning models, predictive analysis is conducted on the health characteristic matrix to identify the health risks that the user may face in the future and provide preventive advice. Health trend analysis and risk prediction can provide a data-driven health management solution for the user, improving the accuracy and forward-looking of health intervention. The generated health information characteristic matrix can not only help to understand the user's current health status but also provide a reference for the user's long-term health trends, helping the user to achieve continuous and effective health management.

[0024] Specifically, in step S3 of the embodiment provided by the present invention, from the health information feature matrix, the data recording time of each smart wearable device is first extracted. For example, the usage duration, usage frequency, usage time period (such as morning, evening), data update frequency of each device, etc. Each column can represent the usage characteristics of the device, and these characteristics include but are not limited to the startup time, stop time, and operation time of the device. Based on the timestamp data of the device, the average usage duration and usage frequency of each device are calculated. These duration data can be obtained through statistical analysis, such as the total usage duration per week, the distribution of each usage duration, etc. The device usage time of the user is segmented. For example, a day is divided into several time periods (such as morning, daytime, evening), and then the usage frequency and duration of the device in these time periods are analyzed to understand the degree of dependence of the user on each device in different time periods. The extracted device usage time characteristics can reflect the degree of dependence and usage habits of the user on each device, help construct the usage pattern of the user, and the extraction of time characteristics provides a basis for subsequent pattern analysis, which can reveal the rules and preferences of device usage. By calculating the device usage duration, frequency, and time segmentation, the specific usage behaviors of the user for each device can be identified, and then the contribution of the device to health management can be analyzed.

[0025] More specifically, based on the device usage time characteristics, clustering analysis (such as clustering algorithms like K-means, DBSCAN, etc.) is applied to classify the usage patterns of the devices. For example, some devices may only be used in the morning (such as a smartwatch monitoring morning heart rate, blood sugar, etc.), while others may be used all day long (such as a sleep monitoring device). Through clustering, the usage rules of the devices can be identified, such as high-frequency usage devices, low-frequency usage devices, dedicated devices, etc. By analyzing information such as the device usage time period and usage frequency, the preferences of the user for each device can be understood. For example, some users may prefer to use a certain health device to monitor their exercise volume, while others may be more dependent on a sleep monitoring device. Analyzing these preferences can provide personalized device usage suggestions for the user.

[0026] More specifically, analyze the usage interaction between different devices to explore whether there is a complementary relationship between the devices. For example, a user may use a smartwatch and sports shoes (gait monitoring device) at the same time during a certain time period, which reflects the collaborative working mode between the devices. Through the interaction analysis between the devices, the comprehensive health management mode of the user can be revealed, and the time pattern of device usage can be explored through time series analysis. For example, the usage of some devices may be periodic (such as using a fitness device once a week), while the usage of others may be continuous (such as a heart rate monitoring device being used all day long). This time series analysis can reveal the rhythm and rules of the user's device usage.

[0027] More specifically, the clustering and summarization of device usage patterns can help identify the device dependence in user health management, reveal the differences between frequently used and infrequently used devices, and through the analysis of device usage preferences, provide personalized health management suggestions for users, optimize device usage strategies, and ensure that users can use the appropriate device at the appropriate time. The analysis of device interaction patterns helps to understand how different devices work together and how to improve the efficiency of user health management through the complementary effects between devices. Time series pattern mining can help capture the regularity of device usage and provide a basis for the planning and optimization of health management.

[0028] More specifically, all the analysis results are integrated to create a map of the user's device usage patterns. This map shows the usage of different devices at different times and frequencies. Through this information, the system can build a personalized device usage model for each user and display the user's device usage patterns in the form of charts or reports for easy viewing by users and health management experts. For example, a heat map can be used to show the high-frequency periods of device usage, or a bar chart can be used to show the usage frequencies of different devices. By feeding back the user's device usage patterns to the user, it can help them adjust their health management strategies. For example, if a user finds that some devices are not fully utilized (such as an infrequently used motion monitoring device), it can be recommended that the user increase the usage frequency of this device or adjust the usage method of the device.

[0029] More specifically, the obtained user device usage patterns can comprehensively reflect the user's behavior habits in health management and provide personalized optimization suggestions based on these behaviors. Visualization and report generation help users intuitively understand their device usage status in health management and provide clear data support for decision-making. Through the feedback mechanism, users can optimize their device usage habits, improve the effect of health management, and ultimately improve the comprehensiveness and accuracy of health monitoring.

[0030] More specifically, based on the analysis results of device usage patterns, the system can recommend adjustment plans for device usage. For example, reduce the usage of inefficient devices or increase the usage frequency of certain devices to ensure that the device usage can maximize the improvement of the user's health monitoring effect. Through the analysis of device usage patterns, personalized health management suggestions are generated. For example, if a user uses a sleep monitoring device less at night, the system can recommend that they increase the collection of night sleep data to improve the management of sleep quality. Through the adjustment and optimization of device usage patterns, users can achieve more efficient device usage in health management, ensure the best match between the device and the user's health needs, and personalized health management suggestions can further improve the user's health monitoring effect and help users achieve better health management goals.

[0031] Specifically, in step S4 of the embodiment provided by the present invention, the usage data of each device in the health information feature matrix is converted into a time series according to the time stamp, ensuring that each time point can reflect the change of the health state. For example, health data such as heart rate, blood sugar, and sleep quality should be collected and recorded at regular time intervals. Through the device usage pattern, combined with the usage period, frequency, etc. of the device, the impact of each device on the health data is identified. For example, if a smart watch is used to monitor exercise volume and heart rate, the usage pattern of the device will directly affect the data sequence of heart rate changes, and thus affect the trend analysis of the health state. There may be missing values or incomplete data in the continuous time series analysis, and these missing data need to be processed. Common processing methods include interpolation method, mean filling, etc., to ensure the continuity and integrity of the data. The health data collected by different devices may be different in time, and these data must be time-synchronized for unified trend analysis. By time-aligning the data of different devices, it can be ensured that the change trends of each health information feature can be compared and analyzed within the same time frame. Through the analysis of the health information feature sequence at the continuous time level, the trend of the user's health data changing over time can be captured, reflecting the dynamic evolution process of the user's health status. Through the mapping of the device usage pattern, the correlation between various health data can be identified, and how each device affects the health state can be understood, laying a foundation for further analysis of the health trend.

[0032] More specifically, select a suitable time series analysis model, such as ARIMA (Autoregressive Integrated Moving Average Model), LSTM (Long Short-Term Memory Network), SVM (Support Vector Machine) and other models, to conduct trend analysis on the user's health status. The ARIMA model is suitable for linear trend analysis and can capture the seasonal, trend and random fluctuation parts in the time series. The LSTM model is suitable for predicting non-linear and complex trends and can learn the complex time-dependent relationship of the health status from historical data. The SVM model is suitable for small sample learning and can conduct trend prediction of the health status in a high-dimensional space.

[0033] More specifically, based on the health information feature matrix, select important features related to the health status (such as heart rate fluctuation, gait, sleep quality, etc.), and construct a feature vector as the input for trend analysis. Use the selected model for trend fitting to obtain the user's health trend curve. For example, the heart rate trend, sleep quality trend, exercise volume trend, etc. of the user can be obtained. These trends can reflect the overall health change of the user. Evaluate the prediction error of the model through methods such as cross-validation, and continuously adjust the model parameters to optimize the health trend prediction effect. For example, evaluate the fitting effect through indicators such as mean square error (MSE).

[0034] More specifically, through trend analysis, it is possible to predict the changes in a user's health condition over a period of time in the future, provide dynamic health trends, and select an appropriate time series analysis model that can adapt to the characteristics of different users' health data, providing accurate health predictions and analyses for users. Health trend prediction can reflect potential changes in health conditions, helping users adjust their lifestyles or take intervention measures in a timely manner.

[0035] More specifically, from the health condition trends obtained by fitting through the above model, extract the key features of the trends (such as the rising rate, falling rate, fluctuation amplitude, etc.). These features can help reveal the changing patterns and development trends of health conditions. Conduct statistical analysis on the fitted trend features to construct the feature distribution of health conditions. For example, probability distribution models (such as normal distribution, exponential distribution, etc.) can be used to describe the changing trends of health conditions. For different health indicators (such as heart rate, blood pressure, sleep quality, etc.), conduct trend fitting and feature distribution extraction respectively. Then, comprehensively analyze the changing trends of multiple health indicators to obtain a global health trend distribution. Display the changes in health trend features through charts (such as line charts, heat maps, etc.), enabling users to clearly understand the evolution process of their own health status. It is also possible to display the stability and change range of health conditions through distribution maps. Through the extraction of the fitted health trend feature distribution, the changing status of users' health can be accurately described, providing a quantitative basis for health management. The health trend distribution can help identify potential risks in health conditions, discover abnormal changes in health conditions, and then provide early warning and intervention suggestions for users. The visual display of health trends enables users to intuitively understand their own health changes, promoting the effectiveness of health management and users' health awareness.

[0036] More specifically, analyze abnormal fluctuations or negative changing trends in health trends, such as too high heart rate, a sharp decline in sleep quality, etc., to identify potential health problems. Based on the changes in health trends, provide personalized health management suggestions for users. For example, if there are abnormal fluctuations in the trend of heart rate, the system may recommend increasing physical activity, adjusting diet, or undergoing more professional health examinations. According to users' feedback and the results of health status monitoring, adjust the health trend prediction model to make health management suggestions more accurate and real-time. Personalized health suggestions and intervention measures help users respond to health problems in a timely manner, reducing health risks. Through trend analysis, the system can dynamically track users' health conditions and provide targeted health management suggestions, improving the controllability of users' health conditions.

[0037] Specifically, in step S5 of the embodiment provided by the present invention, through the health data at the continuous time level, appropriate time series analysis methods (such as ARIMA, LSTM, etc.) are used to fit the trend of health information. The fitting result is the trend change of health information characteristics, which is manifested as the time change trends of different health indicators (such as heart rate, blood sugar, gait, etc.). Key trend characteristics, such as rising rate, falling rate, fluctuation range, seasonality, etc., are extracted from the health trend. These characteristics can reflect the potential change trend of the health condition in the future time.

[0038] More specifically, according to the trend analysis result, a probability distribution model (such as normal distribution, exponential distribution, etc.) is used to model the health trend characteristics, and a distribution model for each health information characteristic is obtained. For example, probability distributions of future values can be established respectively for health indicators such as heart rate, body temperature, blood pressure, etc. Health trend fitting can reveal the laws and trends of different health indicators changing over time. The fitting characteristics of the health trend provide clear time series data and distribution models for subsequent predictive analysis, ensuring the accuracy and reliability of the prediction.

[0039] More specifically, based on the data of health information collection devices (such as smart watches, fitness devices, medical sensors, etc.), the health information at different time points is organized into a health information feature matrix. Each row of this matrix represents the data at a time point, and each column represents a certain health indicator (such as the number of steps, heart rate, sleep quality, etc.). Necessary preprocessing is performed on the health information matrix, such as removing outliers, filling in missing values, standardization, and normalization, etc., to ensure the stability and consistency of the data. For example, interpolation methods can be used to fill in missing data, and the Z-score standardization technique can be used to unify the data to the same dimension. The construction of the health information matrix provides a basic data platform for health trend analysis and prediction, and the preprocessing of the data ensures the reliability of the subsequent analysis results and reduces the interference of abnormal data on model prediction.

[0040] More specifically, according to the structure and data characteristics of the health information feature matrix, an appropriate predictive analysis model is selected. Commonly used models include: the ARIMA model is suitable for processing time series predictions with linear trends, the LSTM (Long Short-Term Memory Network) is applicable to processing health data with long sequences and complex time-dependent relationships, the Support Vector Machine Regression (SVR) is suitable for regression analysis of smaller data sets, and the regression analysis model performs feature prediction through multiple linear regression or non-linear regression, etc. More specifically, it is trained using a known health information feature matrix to learn the patterns and trends in the data. By using historical data to fit the model, the model can accurately reflect the changing trends of health indicators. The prediction performance of the model is evaluated through methods such as cross-validation, and the model parameters are adjusted according to the errors (such as mean squared error, absolute error, etc.) to ensure a high prediction accuracy. Selecting an appropriate prediction model can efficiently capture the time-dependence and trends in health data, providing a reliable basis for predicting health information features. Through training and optimizing the prediction model, accurate prediction of future changes in the user's health status can be achieved.

[0041] More specifically, the trained prediction model is applied to the health trend fitting feature distribution. Based on the existing health data and their trend characteristics, future health information features are predicted. For example, the number of steps, heart rate changes, blood glucose levels, etc. of the user in the next week are predicted. According to the prediction results, a new health information feature matrix is generated, which contains the predicted health indicators and their corresponding time points. By comparing the differences between the actual health data and the predicted data, the model can be continuously adjusted to further improve the prediction accuracy. When predicting health data, considering the prediction errors and uncertainties, confidence intervals or prediction intervals can be calculated to provide a reliable range for the prediction results.

[0042] More specifically, through health prediction based on trend fitting features, accurate prediction of the user's future health status can be achieved, helping the user understand the changing trends of their health conditions. The generated prediction information feature matrix can provide the expected health data at future time points, enabling the user to make appropriate health management decisions based on the prediction information. The confidence intervals and uncertainty analysis of the prediction results enhance the practicality of the prediction, enabling the user to more scientifically assess risks during the health management process.

[0043] More specifically, the predicted health data is visualized through charts (such as line charts, heat maps, scatter plots, etc.) to help the user intuitively understand the health trends. For example, the heart rate change trend and exercise volume change trend in the next few days or weeks are shown. According to the prediction results and the user's actual health data feedback, the prediction model is periodically adjusted and optimized. For example, if there is a large gap between the prediction results and the actual results, the model parameters or data processing methods may need to be adjusted to improve the prediction accuracy. The visualized health prediction information enables the user to more easily understand their health change trends, facilitating timely health adjustments and decisions. Through the feedback mechanism and model adjustment, the prediction analysis can be continuously optimized to improve the accuracy of long-term predictions.

[0044] Specifically, in step S6 of the embodiment provided by the present invention, based on the health information feature matrix and the trained prediction models (such as ARIMA, LSTM, etc.), a health information feature matrix for future time points is generated. This matrix contains predicted health indicators (such as steps, heart rate, blood glucose, etc.) and corresponding time nodes. The generated health information feature matrix will become the basic data for subsequent verification processing. The predicted data usually includes the predicted values of the health status at each time point, representing the health status of the prediction result in the future. This step provides the basis for the subsequent verification and correction processes, ensuring that the expected trends and values of the health data are consistent with the model predictions.

[0045] More specifically, in the future time period, the health data of the user is collected in real time. This data can be obtained through various health monitoring devices (such as smart watches, health sensors, etc.). The real-time data is usually the actual health information and is not affected by the model predictions. The predicted information feature matrix generated for the future time is matched with the actually collected health information, and the differences between the two are compared, especially the deviations between the actual values and the predicted values of each health indicator (such as blood glucose, heart rate, etc.). According to the differences between the prediction results and the actual data, error values (such as mean square error, absolute error, etc.) are calculated. These error values can help evaluate the accuracy of the prediction model. The evaluation indicators include: the deviation between the predicted value and the actual value, and whether the change trend of each health indicator is consistent with the actual trend. Through the verification process, the degree of agreement between the prediction result and the actual health data can be effectively evaluated. The verification process reveals the accuracy of the prediction model and potential prediction biases, and helps to identify and correct the deficiencies of the model.

[0046] More specifically, according to the error results in the verification process, identify which health indicators have large prediction deviations. For example, if the predicted result of the heart rate is too high or too low, or the predicted trend of the blood glucose does not match the actual data, these features need to be focused on. Through error analysis, the parameters in the health trend fitting model are adjusted. Common adjustment methods include: optimizing the parameters of the trend analysis model (such as ARIMA), adjusting the order or smoothing parameters of the prediction model. In machine learning-based methods (such as LSTM, support vector machines, etc.), adjusting hyperparameters such as the learning rate, the number of hidden layers, and the number of nodes of the model. Based on the new adjusted parameters or the corrected model, re-perform trend fitting. Re-fit the health information feature matrix through the adjusted model and generate a new health trend fitting feature distribution. This process enables the health trend model to more accurately reflect the real data.

[0047] More specifically, according to the newly fitted distribution of health trend characteristics, update the prediction intervals and distribution characteristics (such as the range of change, volatility, etc.) of each health indicator, so that the new distribution characteristics are closer to the actual health trend. The feedback correction process can optimize the model according to the verification results, thereby improving the accuracy and reliability of future health information prediction. By adjusting the model parameters, the fitting ability of the model to the health data trend can be enhanced, making future predictions more in line with the actual situation, reducing errors, and improving the predictability of health trends.

[0048] More specifically, through the feedback correction of the health trend fitting feature distribution, a new health trend verification feature distribution is generated. This distribution makes a more accurate prediction of the future health status on the basis of fully considering the actual health data. The new health trend verification feature distribution can be used as the input of the subsequent prediction model to further improve the prediction accuracy. The prediction of the model at future time points will be based on this verification distribution, so as to generate more reliable health predictions in the next cycle. Through continuous verification processing and feedback correction, the model is continuously optimized, making each round of prediction results closer to the actual health data. The system gradually forms a closed-loop prediction and verification mechanism, gradually realizing the precise control of health management. The generated health trend verification feature distribution is more accurate than the initial prediction distribution and reflects the dynamic changes of more actual health data. The continuous update and application of the verification feature distribution enable the health prediction system to adapt to the diversity of individual health changes and enhance the effect of personalized health management.

[0049] More specifically, continuously monitor the user's health data, regularly compare the prediction results with the actual data to form a continuous feedback mechanism. After each prediction cycle, further optimize the prediction model according to the new verification data and feedback results. The model can be continuously iteratively updated according to the new health trend distribution. Based on the updated health trend verification feature distribution, more reasonable health management suggestions are formulated. For example, according to the adjusted blood sugar change trend, it is recommended that the user change their eating habits or increase their exercise amount. The continuous optimization process can improve the adaptive ability of the model. As time goes by, the prediction accuracy is continuously improved. The real-time health management suggestions are adjusted according to the verified health trend, making the user's health management strategy more scientific and personalized.

[0050] Specifically, in step S7 of the embodiment provided by the present invention, health data of users is collected in real time through various health monitoring devices (such as smart watches, health monitors, sensors, etc.). Common health data includes heart rate, number of steps, blood pressure, blood sugar, sleep quality, exercise amount, etc. The collected raw data is cleaned and preprocessed, such as removing noise data, filling in missing values, normalizing / standardizing data, etc. To ensure the consistency and integrity of the data, health data from different time periods and different sources is aggregated to generate a health data set of the user, providing accurate and complete health data and laying a reliable input foundation for subsequent analysis and modeling.

[0051] More specifically, meaningful features are extracted from the health data, such as daily average number of steps, average heart rate, maximum blood sugar value, sleep cycle, etc. These features help to describe the overall health status and trends of the user. The extracted features are organized into a health information feature matrix. The rows of this matrix represent different time points or periods, and the columns represent different health indicators. The features in the matrix are standardized or dimension-reduced (such as PCA, t-SNE, etc.) to facilitate subsequent analysis and visualization processing, generating a structured health information matrix, which is convenient for analyzing health data in multiple dimensions. The processing of standardization and dimension reduction helps to reduce data redundancy and improve the efficiency and accuracy of analysis.

[0052] More specifically, based on historical health data and related features, a health trend fitting model (such as ARIMA, LSTM, etc.) is applied to predict the future health change trends of the user. According to the fitted health trends, the characteristic distribution of each health indicator is generated, describing the possible change range and volatility of this indicator in a future period of time. The fitting model is optimized using historical data and the existing health status to ensure that the prediction of health trends is more accurate. Through the health trend fitting characteristic distribution, the future prediction distribution of each health indicator is obtained, providing a trend basis for subsequent health management. The optimized trend prediction can reflect the actual health changes of an individual and avoid overly rough or deviated health trends.

[0053] More specifically, the actual health data of the user is used for verification, comparing the deviation between the predicted health trend and the actual health status. According to the verification results, the error of the health trend fitting model is corrected, adjusting the health trend characteristic distribution to make it more accurate. According to the corrected trend prediction, the health trend verification characteristic distribution is regenerated to obtain the corrected health data distribution, which is more in line with the actual health data. Through the verification step, the accuracy of the prediction is further improved, ensuring the authenticity and reliability of the health trend. The obtained health trend verification distribution can reflect the real changes in the user's health status and future trends, providing a basis for personalized health advice.

[0054] More specifically, information such as the user's health management data, health information feature matrix, health trend fitting feature distribution, and health trend verification feature distribution is integrated in the same platform or system, and these data are comprehensively analyzed from multiple dimensions such as time, space, individual characteristics, and health indicators. For example, the error between the trend prediction of each health indicator and the actual data is compared, and combined with the historical health records, the changing trend of the user's health status is analyzed. Through multi-dimensional analysis, a health management map of the user is generated. The health management map includes the historical change trend, prediction trend, actual health status, and prediction error of different health indicators, which can reflect the comprehensive information of the individual's health status. By presenting the health management map to the user in a data visualization manner, forms such as charts, dashboards, and heat maps can be used to enable the user to intuitively see their health status and future trends. The multi-dimensional information integration can comprehensively and synthetically analyze the user's health data and trends, providing comprehensive support for health management. The health management map provides an intuitive display of the user's health data, helping the user understand their health status, see the health change trend, and make scientific health decisions.

[0055] More specifically, based on the health management map, the system can comprehensively analyze the user's current health status, prediction trend, actual health data, and error to generate a health feedback report. Combining the health trend verification distribution and the user's health data, the system can provide personalized health suggestions according to the user's health risks and trend changes. For example, if there is an abnormal fluctuation in a certain health indicator, the system can recommend that the user adjust their diet or increase exercise. By combining the user's health feedback with their actual behavior and continuously adjusting the health management map and suggestions through feedback, a closed loop of health management is formed.

[0056] The present invention provides a health management data fusion method based on multiple intelligent wearable devices, which has the following beneficial effects: The present invention obtains the management information of the user's intelligent wearable devices to construct the user's health information collection network, obtains health management data based on the collection network, and uses a health assessment AI model for preliminary evaluation to form a health information feature matrix. The usage time features of each device are extracted to analyze the device usage pattern, and health trend analysis is performed through the health information feature matrix to obtain a health trend fitting feature distribution. Prediction analysis is carried out based on the trend distribution to generate a prediction information feature matrix, and information verification processing is performed. The prediction results are fed back and corrected to form a health trend verification feature distribution. Finally, all health data and feature distributions are integrated to generate a health management map, solving the problem of low data analysis and utilization rate of multiple intelligent wearable devices in the prior art.

[0057] Preferably, the steps of obtaining the management information of the user's smart wearable device and analyzing the relevance of the health management data collection method for the user according to the management information of the smart wearable device to obtain the health information collection network of the user include: S11: Authenticate the account information of the user's smart wearable device through a pre-deployed health data management platform, so that the smart wearable device that has completed the account information authentication is in a data connection state with the health data management platform; S12: Collect the device performance information of the smart wearable device that is in a data connection state with the health data management platform to obtain the device wearing method information, device data collection method information, and device sensing performance information of the smart wearable device. The device wearing method information, device data collection method information, and device sensing performance information of each smart wearable device together constitute the management information of the user's smart wearable device; S13: Simulate the process of collecting user information for each smart wearable device according to the management information of the smart wearable device to obtain the user information collection simulation characteristics of each smart wearable device, and analyze the tendency of the user health feedback value for the user information collection simulation characteristics of each smart wearable device according to the management information of the smart wearable device to obtain the health feedback pointing distribution of each smart wearable device; S14: Based on the user information collection simulation characteristics and health feedback pointing distribution of each smart wearable device, perform digital simulation of the user health data management form for each smart wearable device respectively to obtain the device simulation unit corresponding to each smart wearable device; S15: Analyze the relevance of the device simulation units of each smart wearable device, and configure the parameters of node deployment and link connection for each device simulation unit based on the results of the relevance analysis, so that each device simulation unit is converted into a health information collection node interconnected by information association links to construct the health information collection network of the user.

[0058] Specifically, when authenticating the account information of the user's smart wearable device through the health data management platform, each smart wearable device needs to provide the user's account information (such as device ID, user identity authentication, etc.), and verify the identity through the platform to ensure that the device is legal and can access the platform. After successful authentication, the smart wearable device establishes a data connection with the health data management platform to ensure that the device can start real-time transmission of health data, ensure the security and reliability of data exchange between the device and the platform, prevent unauthorized devices from accessing the platform, protect data privacy and security. Device authentication enables the platform to synchronize data with legal smart devices, providing a reliable data source for subsequent health data analysis.

[0059] More specifically, performance information of smart wearable devices that are in a data connection state with the health data management platform is collected, specifically including the following information: device wearing method information, such as the wearing position of the device (wristband, chest strap, earphone, etc.) and the wearing method (whether it fits closely to the skin, whether it is interfered, etc.), device data collection method information, how the device collects health data (e.g., sensor type, sampling frequency, etc.), device sensing performance information, the sensing accuracy of the device (such as heart rate monitoring accuracy, step counting accuracy, etc.). These device information are collected and stored to form the management information of the user's smart wearable device, providing basic data for analyzing the device performance and evaluating its health data collection ability, helping to judge the adaptability and accuracy of the device, understanding the performance characteristics of different devices, contributing to optimizing the quality of health data collection, and providing personalized health management suggestions.

[0060] More specifically, based on the management information of each smart wearable device, the device is simulated to predict the health data collection effect of the device under different wearing methods, data collection methods, and sensing performances. By simulating and analyzing the user information collection characteristics of each device, the "user information collection simulation characteristics" of the device under specific conditions are obtained, that is, the health data capture ability and effect of the device in different usage scenarios. The simulation process can predict the performance of the device in actual use, thus providing a more accurate prediction model for subsequent health feedback and data analysis, helping to evaluate the applicability and accuracy of the device, and finding the most suitable device and wearing method for specific users.

[0061] More specifically, based on the user information collection simulation characteristics of the device, a tendency analysis of the device's health feedback is conducted. Specifically, analyze how the data collected by each smart wearable device affects the user's health management feedback (for example, whether potential health risks can be detected in a timely manner, whether effective health improvement suggestions can be provided, etc.). By analyzing the collection characteristics of different devices, the health feedback pointing distribution is obtained, which describes the types and priorities of health feedback that the device can generate (for example, some devices focus on heart rate monitoring, and some devices focus on sleep quality). A health feedback tendency analysis based on simulation characteristics is provided, enabling the health feedback of each device to match the actual health needs, helping to improve the personalization and accuracy of the health management system, and enabling users to obtain more targeted health suggestions.

[0062] More specifically, based on the user information collection simulation characteristics and health feedback pointing distributions of each smart wearable device, digital simulations are performed on each device to simulate its actual application in user health data management. Through digital simulations, a "device simulation unit" for each device is created. Each device simulation unit represents the role and function of the device in user health data management, and simulates how it affects the user's health data feedback. Digital simulations help to intuitively display the effects of devices in health data management, simulate the impacts of different devices and wearing methods on health management, provide in-depth understanding of device performance, and assist in evaluating the advantages and limitations of different devices in data collection and feedback.

[0063] More specifically, through the correlation analysis of device simulation units, the interrelationships between different devices are studied. For example, when a user wears multiple devices, how they work together, how data is interconnected, and how feedback is complementary. According to the results of the correlation analysis, parameter configurations for node deployment and link connectivity are performed on the device simulation units to ensure data exchange and interconnection between different device simulation units through information association links, enabling the device simulation units to be interconnected to form a health information collection network with information flow. Through correlation analysis, multiple devices can be effectively integrated into a collaborative health information collection network, optimizing the health data collection and management processes, ensuring the coordinated operation of each device in the user health management system, and improving the coverage and accuracy of health data.

[0064] More specifically, each device simulation unit is transformed into interconnected health information collection nodes, ultimately forming the user's health information collection network. This network enables the comprehensive collection of health data through node connections between smart wearable devices. The health information collection network is continuously optimized to ensure that each node can collect and transmit data in real time and accurately. The constructed health information collection network can comprehensively and real-time capture the user's health data, realize data sharing and collaborative management among multiple devices, improve the breadth and depth of user health data collection, and provide a solid data foundation for subsequent health analysis, risk warning, and personalized health management.

[0065] Preferably, the steps of obtaining the user's health management data based on the user's health information collection network and performing preliminary evaluation and information conversion on the health management data according to a pre-trained health assessment AI model to obtain the user's health information feature matrix include: S21: Information collection is performed on the user through the smart wearable device to obtain the original detection data collected by the smart wearable device, and the device label and time label of the original detection data are synchronously generated; S22: Based on the health information collection network, real-time receive the original detection data with device tags and time tags, and based on the health information collection network, locate the health feedback value of the original detection data with device tags and time tags to obtain the data feedback value characteristics of the original detection data; S23: Configure the data management characteristics of the original detection data according to the data feedback value characteristics to obtain the health management data corresponding to the original detection data; S24: Retrieve the corresponding health assessment AI model from the model deployment layer in information connection state with the health information collection network according to the health management data, and through the health assessment AI model, conduct a feedback assessment of health information on the health management data to obtain the feedback assessment information of the health management data; S26: Convert the vector expression format of the feedback assessment information to obtain the health information feature vector corresponding to the feedback assessment information, and conduct a structural analysis of the device features and time features on the health information feature vector according to the health management data corresponding to the feedback assessment information to obtain the correlation structure features between the health information feature vectors; S26: Construct a matrix for each of the health information feature vectors according to the correlation structure features to obtain the health information feature matrix of the user.

[0066] Specifically, information of the user is collected through smart wearable devices to obtain the user's health data (such as heart rate, blood pressure, steps, body temperature, etc.). The original data collected by the device not only includes health data, but also synchronously generates the device tag (identifying the data source device) and time tag (recording the data collection time) for each piece of data, ensuring the integrity and traceability of the data. The device tag and time tag provide source and timeliness information for the data, which helps subsequent data analysis and processing, provides real-time health data with timeliness for the health management system, and makes data collection more accurate and orderly.

[0067] More specifically, based on the health information collection network, real-time receive the original detection data with device tags and time tags. Analyze the data through the health information collection network, match each piece of original detection data with a preset health feedback model, locate the health feedback value characteristics of each piece of data, realize real-time data access and analysis, ensure fast data feedback and timely processing, provide a preliminary health value assessment of the data, and can adjust the management strategy according to the user's health condition and give personalized health guidance.

[0068] More specifically, according to the health feedback value characteristics of each piece of original detection data, the data management characteristics are configured. For example, the data is prioritized according to signal quality, acquisition accuracy, and feedback value, or the data is classified according to the level of health risk. According to the configured data management characteristics, the original data is cleaned, preprocessed, and structurally stored to form health management data, ensuring the quality and usability of the data. By preprocessing the data, subsequent analysis becomes more efficient and accurate, improving the data processing ability and flexibility, and providing accurate and timely health management suggestions for users.

[0069] More specifically, the pre-trained health assessment AI model is retrieved from the model deployment layer of the health information collection network. The health management data is feedback-evaluated through the health assessment AI model to generate corresponding health assessment information. The AI model usually analyzes, predicts, and evaluates health data based on machine learning and data training. Through the health assessment AI model, intelligent analysis of health management data is achieved, automatically evaluating the user's health status and providing objective feedback. The AI model can identify potential health risks and give early warnings to help users intervene or improve early.

[0070] More specifically, the feedback evaluation information output by the health assessment model is converted into vector representation. The feedback evaluation information (such as health risk, health status, etc.) will be converted into a vector in a specific format for subsequent processing and analysis. Converting the complex health assessment results into structured data (vectors) makes subsequent analysis and processing more efficient and executable. The vectorized health information can provide a basis for subsequent correlation analysis and feature mining.

[0071] More specifically, based on the health management data corresponding to the feedback evaluation information, a structural analysis of device characteristics and time characteristics is carried out. Specifically, the impact of device characteristics (such as device type, wearing position, etc.) and time characteristics (such as the time period and frequency of data acquisition) on health data is analyzed. To gain a deeper understanding of the impact of device performance and data acquisition time, the structural analysis helps to discover potential patterns in health data, optimize health data management, and better adapt to personalized health assessment needs by analyzing time and device characteristics.

[0072] More specifically, according to the analysis results of device characteristics and time characteristics, the associated structural characteristics between health information feature vectors are extracted. This process will reveal the interaction between different devices and time characteristics and how they affect the feedback evaluation of health data. The extracted associated structural characteristics help to establish a clear health data network, revealing the complex relationships between different variables (such as device type, wearing method, time, etc.). The associated structure analysis helps to further improve the intelligent processing ability of health management data and provide accurate health feedback.

[0073] More specifically, based on the extracted associated structural features, a health information feature matrix is constructed. This matrix integrates the relationships between different health data, providing multi-dimensional data support for the comprehensive analysis of the user's health status. The health information feature matrix provides a comprehensive and systematic data view, which can fully reflect the user's health status and is helpful for complex data analysis, such as health trend prediction and personalized health intervention. Ultimately, it provides reliable data support for the user's long-term health management.

[0074] Preferably, the steps of extracting the usage time features and summarizing the patterns of each smart wearable device according to the health information feature matrix to obtain the device usage patterns of the user for each smart wearable device include: S31: Extract the features of the single-device usage time distribution and the overall device usage time correlation of each health information feature vector according to the health information feature matrix to obtain the single-device usage time distribution feature sequence and the overall device usage time correlation feature sequence of the health information feature matrix; S32: Based on the single-device usage time distribution feature sequence of each smart wearable device and the overall device usage time correlation feature sequence, perform feature analysis of device usage trend, device usage periodicity, and device usage volatility, and perform weighted fusion processing on the analyzed device usage trend features, device usage cycle features, and device usage fluctuation features to obtain the information collection pattern of the health information feature matrix; S33: Perform pattern matching processing on the information collection pattern according to several preset usage patterns stored in the reserve database to obtain the matching indexes of the information collection pattern and various preset usage patterns; S34: According to the matching indexes of the information collection pattern and various preset usage patterns, perform pattern weighted fusion processing on various preset usage patterns in the reserve database to obtain the device usage patterns of the user for each smart wearable device.

[0075] Specifically, according to the health information feature matrix, extract the usage time distribution features of each smart wearable device, that is, for each device, record the specific time, frequency, and duration of its use. For the overall device usage time correlation feature extraction, extract an overall device usage time correlation feature from the data of all smart wearable devices, that is, analyze the relationship between the usage times of all devices. For example, the correlation between the usage frequency of a certain device and the usage frequencies of other devices. By analyzing the time distribution of each device and the relationships between devices, we can more comprehensively understand the usage behaviors and patterns of devices. Through the analysis of the health information feature matrix, form the correlation between the time features of each device and the overall usage time, which helps to extract more valuable device usage patterns, can clearly show the usage situation of each device by the user, and lay a foundation for further behavior pattern analysis.

[0076] More specifically, analyze the usage trend of each device within a certain time period, such as whether the device usage amount is gradually increasing or decreasing. Trend analysis can reveal the growth or decline trend of device usage. Analyze the periodic characteristics of device usage, and explore whether users tend to use a certain device during specific time periods (such as morning, evening). Analyze the volatility of device usage time, that is, whether there are frequent fluctuations when users use the device, such as the irregularity of usage time. Weightedly fuse the device usage trend features, periodic features, and volatility features obtained from the above analysis to form a comprehensive information collection mode. Through the comprehensive analysis of features such as trend, period, and volatility, we can more accurately identify the device usage behavior patterns of users. For example, some devices may have a trend of high - peak usage, while other devices may be in a relatively stable usage state. Through the weighted fusion method, we can comprehensively consider various device usage features, making the final obtained health information collection mode more accurate and comprehensive, and capable of reflecting the real user behavior.

[0077] More specifically, extract several preset usage patterns from the reserve database. These patterns may be obtained based on big data analysis, industry experience, or historical data training. Match the information collection mode obtained from the health information feature matrix with the preset usage patterns in the reserve database. By comparing the similarity between the actually collected usage pattern and the preset pattern, calculate the matching index. By comparing the differences between each preset pattern and the current information collection mode, obtain the matching index. The matching index reflects the similarity between a user's device usage pattern and a preset pattern. The higher the value, the closer the user's device usage pattern is to a certain preset pattern. The calculation of the matching index can help the system automatically identify the similarity between the user's device usage pattern and the preset pattern, providing a basis for applications such as personalized recommendation and health intervention. Through precise matching and index calculation, the system can provide users with more accurate health management solutions, helping device manufacturers optimize product designs or service strategies.

[0078] More specifically, based on the matching index, a weighted fusion process is performed on the preset usage patterns. That is, according to the level of the matching index and combined with the characteristics of each preset pattern, a comprehensive device usage pattern is calculated. Through the weighted-fused pattern, the user's final device usage pattern is obtained. This pattern not only reflects the user's usage characteristics of each smart wearable device but also reveals the user's health management needs and device usage preferences. The finally obtained device usage pattern is more personalized, can accurately reflect the user's usage habits and needs on different devices, provides an accurate basis for subsequent health interventions, device optimizations, etc. Through the weighted fusion method, the system can provide more accurate usage suggestions and health management solutions for the user, further improving the user experience and health effects of smart wearable devices.

[0079] Preferably, the step of performing a trend analysis of the health status at the continuous time level on the health information feature matrix according to the device usage pattern to obtain the health trend fitting feature distribution of the user includes: S41: Analyze the health display form of the user according to the device usage pattern to obtain the state display criteria of several possible health states of the user corresponding to the device usage pattern at the continuous time level, and combine the state display criteria of various possible health states of the user corresponding to the device usage pattern at the continuous time level to obtain the continuous time evaluation criteria of the health information feature matrix corresponding to the device usage pattern; S42: Perform continuous time-level feature correlation fitting verification on the health information feature matrix respectively according to the various state display criteria in the continuous time evaluation criteria to obtain the health state standard fitting curves of the health information feature matrix corresponding to various state display criteria; S43: Extract multi-level fitting features from the health state standard fitting curves of the health information feature matrix corresponding to various state display criteria to obtain the highly fitting part, ordinary fitting part, and abnormal fitting part of each health state standard fitting curve; S44: Evaluate the confidence level of each health state standard according to the distribution ratio of the highly fitting part, ordinary fitting part, and abnormal fitting part of each health state standard fitting curve to obtain the state confidence level of the possible health state corresponding to each health state standard; S45: Analyze the overall fitting method of each possible health state according to the state confidence level of each possible health state, and extract and combine the health trend tendencies of each possible health state according to the overall fitting method to obtain the health trend fitting feature distribution of the user.

[0080] Specifically, analyze the user's health status according to the device usage pattern to determine the display form of the user's health status at different time points. For example, the display of the health status can be based on the change trend of physiological parameters (such as heart rate, blood pressure, etc.) recorded by the smart wearable device. According to different health statuses, identify and define the related device usage patterns. This may include indicators such as usage time, frequency, volatility, etc. Define a status display standard at the continuous time level for each health status, that is, how the relationship between the health status and the device usage pattern is displayed at a specific time point or time period. The display standard may include characteristics such as the fluctuation range, stability, and mutation of the status. By analyzing the relationship between the user's device usage pattern and the health status in detail, a basis for subsequent health trend analysis can be established. This provides clear criteria and a framework for accurately judging the user's health status. By establishing a standardized status display form, it helps to systematically analyze and evaluate the change trends of different health statuses.

[0081] More specifically, according to the health display standard, process the health information feature matrix using the evaluation criteria at the continuous time level. The evaluation criteria are based on the continuous status display of time series data, mainly evaluating the matching degree between the device usage pattern and the change of the health status. Conduct a fitting verification of the health information feature matrix at the continuous time level, that is, examine the correlation between the health status and the device usage pattern on the time axis. Through the generation of the fitting curve, verify whether the change of the health status in different time periods conforms to the expected pattern. Through the fitting verification at the continuous time level, the system can more accurately track the change trend of the user's health status over time. Through the fitting verification, the dynamic relationship between the device usage pattern and the health status can be quantified, providing data support for health management decisions.

[0082] More specifically, extract the highly fitting part from the fitting curve, that is, the part that best matches the health status standard, representing the typical trend of the health status. Extract the part with a general matching degree to the health status standard, representing the fluctuation or stability of the health status. Identify the part of the fitting curve that is significantly inconsistent with the health status standard, which may reflect abnormal fluctuations or mutations in the health status. Through multi-level fitting feature extraction, different levels of manifestations of the health status (such as normal fluctuations, trend changes, abnormal fluctuations) can be distinguished, providing more-dimensional analysis for subsequent health status prediction. Extracting the abnormal fitting part helps to identify potential health risks or sudden situations, providing timely information support for health warnings.

[0083] More specifically, the distribution ratios of different parts of the health status standard fitting curve (highly fitting part, ordinary fitting part, abnormal fitting part) are calculated. Specifically, by analyzing the ratios of these parts in the total fitting curve, the stability of the health status is evaluated. According to the distribution ratios of different fitting parts, the confidence levels of each health status standard are calculated. A high confidence level indicates that the health status is highly consistent with the device usage pattern, while a low confidence level indicates a greater degree of uncertainty in the health status. The confidence level assessment provides a quantitative credibility value for each health status, enabling the system to more accurately judge the reliability of the user's health status. Through the confidence level assessment, a more reliable basis for health prediction and intervention can be provided for the health management system, improving the accuracy of health management.

[0084] More specifically, based on the confidence levels of the health status standards, a fitting analysis of the overall health status is performed, that is, the changing trends of multiple health statuses are comprehensively considered to analyze their global health changing trends, and the health trend tendencies in the overall fitting result are extracted, that is, the main trends of the user's health status changing over time (such as gradually improving, gradually deteriorating, or remaining stable). Combining the confidence levels of each health status, the future health development trends can be predicted more accurately. Through the overall fitting analysis, the future health trends of the user can be predicted, and forward-looking guidance can be provided for health management. Based on the extraction of the health trend tendencies, the system can provide personalized health suggestions for the user to help them formulate reasonable health improvement or maintenance strategies.

[0085] Preferably, the steps of performing predictive analysis on the health information feature matrix based on the health trend fitting feature distribution to obtain a predictive information feature matrix include: S51: Selecting trend key nodes for the health information feature matrix based on the health trend fitting feature distribution, so as to select trend key nodes from each health information feature vector of the health information feature matrix, and performing node linking on each of the trend key nodes to obtain a predictive starting entity; S52: Simulating the trend development of the predictive starting entity for a future time period according to the health trend fitting feature distribution to obtain the simulated state of the predictive starting entity corresponding to the health trend fitting feature distribution; S53: Analyzing the information interaction relationship between the predictive starting entity and the health information feature matrix, and performing information expansion analysis on the simulated state based on the information interaction relationship to obtain a predictive information feature matrix.

[0086] Specifically, analyze the distribution of health trend fitting features to determine which features are important factors affecting health trends. These factors can be physical sign data, device usage patterns, health changes, etc. Based on each health information feature vector in the health information feature matrix, select the key nodes that have the greatest impact on the health status trend. The key nodes may be certain specific health indicators, change points in device usage patterns, or turning points in the fluctuations of the health status. Link the selected trend key nodes to construct the initial state of the trend development. The links between nodes can be time correlation, state correlation, or other correlations, forming a continuous development chain. By selecting trend key nodes, the core factors affecting health trends can be accurately identified, making the initial stage of predictive analysis more accurate. By only focusing on key nodes, the complexity of the model can be effectively simplified, unnecessary data redundancy can be avoided, and resources can be concentrated for key analysis.

[0087] More specifically, according to the distribution of health trend fitting features, simulate the trend development of the prediction starting entity (formed by linking the selected trend key nodes) for a future time period. This simulation will be based on existing health trend data and a prediction model to generate the trajectory of the health status change of the prediction entity within the future time period. Through the simulated state of the health trend fitting features, predict the change of the health status of the prediction starting entity over time, including possible fluctuations, change trends, stability, etc. Through trend development simulation, a relatively accurate prediction of the future health status can be made, providing predictive data support for health management and decision-making. Simulating the health change trends in different time periods can help identify possible future health risk points and provide dynamic predictions.

[0088] More specifically, analyze the interaction relationship between the prediction starting entity and other information in the health information feature matrix. Specifically, analyze the mutual influence between different health information features, the interaction in time, and the correlation between the device usage pattern and the health status. Based on these interaction relationships, conduct an extended analysis of the simulated state to consider possible influencing factors of health changes. The extended analysis not only includes the change trend of the health status but also comprehensively considers the impact of device usage patterns, environmental factors, user behavior, etc. on the health status. By analyzing the information interaction relationship, systematically understand the multi-dimensional factors of health status changes, avoid being misled by a single factor, and conduct an extended analysis of the simulated state to more comprehensively reflect the diversity of the health status and provide more dimensional information support for prediction.

[0089] More specifically, through the extended analysis of the simulated state, a complete prediction information feature matrix is obtained. This matrix contains the prediction results of the user's future health status and related health trend information. The prediction information feature matrix not only includes the main trends of the health status but also can cover the changes of various health factors (such as physiological data, device usage, environmental data, etc.). Through the extended analysis, the prediction information feature matrix can comprehensively reflect the possible development paths of the future health status. It not only provides the prediction results of the health trends but also helps users or systems make better health management decisions. Through the interactive analysis with real-time data, a dynamic and highly accurate health prediction model can be generated to ensure the timeliness and accuracy of the prediction results.

[0090] Preferably, the step of performing information verification processing on the health information feature matrix generated at a future time according to the prediction information feature matrix, and performing feedback correction on the health trend fitting feature distribution according to the verification result to obtain the health trend verification feature distribution includes: S61: After obtaining the health information feature matrix generated at a future time, extract the substantial abnormal feature of the health information feature matrix according to the prediction information feature matrix to obtain the matrix deviation tendency distribution between the health information feature matrix and the prediction information feature matrix; S62: Perform the possibility feedback analysis of the health status on the matrix deviation tendency distribution according to each possible health status and the overall fitting method corresponding to the health trend fitting feature distribution to obtain the state confidence adjustment index of each possible health status reflected by the matrix deviation tendency distribution; S63: Correct the overall fitting method of the health trend fitting feature distribution according to the state confidence adjustment index of each possible health status reflected by the matrix deviation tendency distribution, and extract and combine the health trend tendencies of each possible health status according to the corrected overall fitting method to obtain the health trend verification feature distribution of the user.

[0091] Specifically, through the health information feature matrix generated at a future time, the health data of the user within a certain period in the future is obtained. These data will be used to verify the accuracy of the prediction results. Analyze the health information feature matrix generated in the future to identify the differences between it and the predicted information feature matrix. These differences usually manifest as "abnormal features", that is, health data that does not conform to the expected or predicted results. Abnormal features may be sudden changes or unexpected fluctuations in the health status, etc. According to the abnormal features, calculate the deviation between the health information feature matrix and the predicted information feature matrix. The distribution of the deviation can reveal the change trend of health information and the difference between prediction and reality. By extracting substantial abnormal features, the deviations and abnormal points in the data can be discovered in a timely manner, providing key clues for further analysis. Through the calculation of the matrix deviation tendency distribution, the differences between health data and predicted data can be systematically quantified, and the specific deviation direction and trend can be clarified.

[0092] More specifically, according to the health trend fitting feature distribution, for the matrix deviation tendency distribution, conduct a likelihood feedback analysis of the health status. This step is based on historical data and trend fitting models to speculate which possible health states will be affected in the change of the health status. Through the feedback analysis of the matrix deviation, calculate the probabilities of the occurrence of each possible health state. The occurrence probability of each health state reflects the possibility of the change of the health state under given conditions. Through the feedback analysis of the matrix deviation, the transition between different health states can be predicted in more detail. Especially for the health states with non-linear changes, sudden changes can be more foreseen. According to the deviation feedback, the health state prediction model can be dynamically adjusted to make it closer to the real situation.

[0093] More specifically, according to the results of the feedback analysis, assign a "confidence adjustment index" to each possible health state. This index represents the credibility of each health state within the future time period. A health state with a high confidence indicates a greater possibility of its occurrence, and vice versa. Correct the confidence of each possible health state in the prediction model to make the prediction results more in line with the actual observed data change trend. Through confidence adjustment, the reliability of the prediction results can be enhanced, and the prediction deviation caused by data errors can be reduced. The confidence adjustment index can provide information about the credibility of each health state, thus making the health prediction model more accurate.

[0094] More specifically, the confidence level is adjusted to modify the overall fitting method in the health trend fitting feature distribution. This modification step takes into account the differences between the real data and the prediction results, making the health trend model closer to the actual situation. Based on the modified health trend fitting feature distribution, the health trend tendencies of each health state are extracted. The health trend tendency describes how each health state evolves over time, thereby further predicting the user's health trend. Combining the health trend tendencies of each health state, the final health trend verification feature distribution is constructed. By modifying the health trend fitting feature distribution, the prediction model can better reflect the actual health change trend and enhance its accuracy. By extracting the health trend tendency, it is possible to better predict the user's future health changes and detect potential health problems in a timely manner.

[0095] More specifically, through the above-mentioned modification and adjustment, the final health trend verification feature distribution is obtained. This distribution not only includes the prediction of the health state, but also considers the possible change trends and the credibility of each health state. Combining multi-dimensional data, a comprehensive analysis of the user's health state is carried out, and personalized health trend prediction and management suggestions are provided for the user. Through the health trend verification feature distribution, more personalized and accurate health predictions can be provided for the user to help them better manage their health. This method can continuously update the user's health trend prediction based on real-time health data and trend changes, making health management more intelligent and dynamic.

[0096] Preferably, the steps of performing multi-dimensional information integration on the health management data, the health information feature matrix, the health trend fitting feature distribution, and the health trend verification feature distribution to obtain a health management map for providing feedback on the user's health data include: S71: Perform data compression and storage processing on the health management data, the health information feature matrix, the health trend fitting feature distribution, and the health trend verification feature distribution to obtain a data record level; S72: Continuously update and record the health trend fitting feature distribution and the health trend verification feature distribution to obtain the health trend verification feature distribution updated in each time period, and perform an overall comprehensive analysis and health state change analysis on the health trend verification feature distribution updated in each time period to obtain the overall health assessment conclusion and the health state change assessment conclusion of the user. Integrate the overall health assessment conclusion and the health state change assessment conclusion to obtain a fusion assessment level; S73: Analyze the user health state certainty detection method for each of the smart wearable devices based on the overall health assessment conclusion and the health state change assessment conclusion to obtain the detection execution plan for each of the smart wearable devices, and construct a guidance interaction interface according to the detection execution plan; S74: Combine the data record level, the fusion evaluation level, and the guidance interaction interface to obtain the health management map.

[0097] Specifically, perform compression processing on health management data, health information feature matrices, health trend fitting feature distributions, and health trend verification feature distributions. Compress these health-related data to reduce redundant data and lower the storage space requirements. Compression methods can include techniques such as data dimensionality reduction, principal component analysis (PCA), and clustering analysis to ensure that important health information is retained while improving data processing efficiency. Store the compressed data hierarchically according to time, health status, or other relevant dimensions to form the "data record level". This level can be organized according to different health indicators, time periods, or device categories, making the data more structured and easier to query. Through data compression and hierarchical storage, the storage space requirements are greatly reduced, and the efficiency of subsequent data querying and analysis is improved. Organizing the data hierarchically for storage facilitates the quick access and analysis of data for specific dimensions or time periods.

[0098] More specifically, regularly update the health trend fitting feature distribution and the health trend verification feature distribution. These distributions represent the changing trends of the user's health status and will be continuously adjusted and corrected over time. Conduct an overall comprehensive analysis of the updated health trend verification feature distribution, and combine the change data within each time period to perform an analysis of the health status changes. This analysis helps identify the user's health trends, the direction of change, possible health risks, etc. By continuously updating the health trend data, it can timely reflect the changes in the user's health status and provide dynamic health management services. Through the analysis of health status changes, it can provide detailed health trend insights for the user to help the user understand the background and reasons for their health changes.

[0099] More specifically, in combination with the updates of the health trend fitting feature distribution and the health trend verification feature distribution, conduct an overall health assessment and an assessment of health status changes. The overall health assessment conclusion is based on the user's long-term health trends, while the health status change assessment conclusion focuses on short-term health fluctuations and changes. Integrate the information of these two assessment conclusions to obtain a comprehensive "fusion evaluation level". This level can comprehensively describe the user's health status, the trend of health changes, and possible health risks. By combining the overall health assessment and the assessment of health status changes, it is possible to comprehensively understand the user's health status, not only identify long-term trends but also reveal short-term changes and potential health problems, providing a clear health status map for health managers and users and facilitating the formulation of more precise health management strategies.

[0100] More specifically, based on the overall health assessment conclusion and the health status change assessment conclusion of the fusion assessment level, analyze the health status confirmation detection method of the smart wearable device used by the user. This analysis aims to determine whether the device can accurately monitor the health status change of the user and propose corresponding optimization solutions. According to the analysis results, design appropriate detection execution plans for each smart wearable device. These plans can involve the usage method of the device, sensor settings, data collection frequency, etc. According to the confirmation detection analysis results, the usage method and data collection strategy of the smart wearable device can be optimized, improving the health monitoring accuracy of the device. By specifically analyzing the performance of the smart wearable device, a customized health monitoring plan can be provided for each user, enhancing the user's health management experience.

[0101] More specifically, based on the detection execution plan, construct a guiding interaction interface. This interface provides information such as the display of health monitoring data, health advice, and device usage guidance for the user. The interface design should be simple and intuitive to ensure that the user can easily understand the health status change, the usage method of the monitoring device, and personal health management advice. Through the friendly interaction interface, the user can conveniently view and understand their own health status, and take timely health management measures. By providing clear feedback and guidance, the user is more motivated to conduct health management and increases their trust in health management tools.

[0102] More specifically, combine the health management data stored in the data record level with the integrated health assessment results (fusion assessment level). In this way, the data record and the assessment results can be displayed integrally, facilitating subsequent analysis and feedback. Combine the guiding interaction interface with the health assessment results to create a final health management map. This map shows multi-dimensional information such as the user's health trend, health assessment conclusion, and the detection plan of the smart device, providing a comprehensive health management service for the user. The health management map can intuitively display the user's health data, trend changes, device usage suggestions, etc., becoming a centralized display platform for the user's health management. Through multi-dimensional data integration, the health management map can provide more accurate and personalized health management advice to help the user make healthier choices in daily life.

[0103] In the second aspect, the present invention provides a health management data fusion system based on multiple smart wearable devices for implementing a health management data fusion method based on multiple smart wearable devices according to any one of the first aspect.

[0104] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A health management data fusion method based on multiple smart wearable devices, characterized in that: include: Acquire the user's smart wearable device management information, and perform a correlation analysis on the user's health management data collection method based on the smart wearable device management information to obtain the user's health information collection network; Acquire the user's health management data based on the user's health information collection network, and perform preliminary evaluation and information conversion on the health management data according to a pre-trained health assessment AI model to obtain the user's health information feature matrix; Extracting usage time features and summarizing patterns of each smart wearable device according to the health information feature matrix to obtain device usage patterns of each smart wearable device by the user; Performing a health status trend analysis on the health information feature matrix at a continuous time level according to the device usage pattern to obtain a health trend fitting feature distribution of the user; Performing a prediction analysis on the health information feature matrix based on the health trend fitting feature distribution to obtain a prediction information feature matrix; Performing information verification processing on the health information feature matrix generated at a future time according to the predicted information feature matrix, so as to feedback and correct the health trend fitting feature distribution according to the verification result, so as to obtain the health trend verification feature distribution; The health management data, the health information feature matrix, the health trend fitting feature distribution and the health trend verification feature distribution are integrated in multiple dimensions to obtain a health management map for providing feedback on the user's health data.

2. The health management data fusion method based on multiple smart wearable devices as claimed in claim 1, characterized in that: The steps of obtaining the user's smart wearable device management information and performing a correlation analysis on the user's health management data collection method according to the smart wearable device management information to obtain the user's health information collection network include: The user's smart wearable device is authenticated through a pre-deployed health data management platform, so that the smart wearable device that has completed the account information authentication is in a data connection state with the health data management platform; Collecting device performance information of smart wearable devices that are in data connection with the health data management platform to obtain device wearing mode information, device data collection mode information, and device perception performance information of the smart wearable devices, wherein the device wearing mode information, device data collection mode information, and device perception performance information of each smart wearable device together constitute the user's smart wearable device management information; Simulate the process of collecting user information for each of the smart wearable devices according to the smart wearable device management information to obtain the user information collection simulation characteristics of each of the smart wearable devices, and perform a tendency analysis of the user health feedback value on the user information collection simulation characteristics of each of the smart wearable devices according to the smart wearable device management information to obtain the health feedback orientation distribution of each of the smart wearable devices; Based on the user information collection simulation characteristics and health feedback orientation distribution of each of the smart wearable devices, digital simulation of user health data management is performed on each of the smart wearable devices to obtain device simulation units corresponding to each of the smart wearable devices; A correlation analysis is performed on the device simulation units of each of the smart wearable devices, and based on the results of the correlation analysis, node deployment and link connectivity parameter configuration are performed on each of the device simulation units, so that each of the device simulation units is converted into a health information collection node that is interconnected by information correlation links, so as to construct a health information collection network for the user.

3. The health management data fusion method based on multiple smart wearable devices as claimed in claim 1, characterized in that: The steps of acquiring the user's health management data based on the user's health information collection network, and performing preliminary evaluation and information conversion on the health management data according to a pre-trained health assessment AI model to obtain the user's health information feature matrix include: Collecting information from the user through the smart wearable device to obtain original detection data collected by the smart wearable device, and synchronously generating a device tag and a time tag for the original detection data; Receiving original detection data with device tags and time tags in real time based on the health information collection network, and performing health feedback value positioning on the original detection data with device tags and time tags based on the health information collection network to obtain data feedback value characteristics of the original detection data; Performing data management feature configuration on the original detection data according to the data feedback value feature to obtain health management data corresponding to the original detection data; According to the health management data, a corresponding health assessment AI model is retrieved from a model deployment layer that is in information communication with the health information collection network, and feedback evaluation of health information is performed on the health management data through the health assessment AI model to obtain feedback evaluation information of the health management data; Convert the feedback evaluation information into a vector expression format to obtain a health information feature vector corresponding to the feedback evaluation information, and perform a structural analysis of device features and time features on the health information feature vector according to the health management data corresponding to the feedback evaluation information to obtain correlation structural features between the health information feature vectors; A matrix is ​​constructed for each of the health information feature vectors according to the association structure feature to obtain a health information feature matrix for the user.

4. The health management data fusion method based on multiple smart wearable devices as claimed in claim 3 is characterized in that: The steps of extracting usage time features and summarizing patterns of each smart wearable device according to the health information feature matrix to obtain the device usage patterns of each smart wearable device by the user include: Extracting the characteristics of the single device usage time distribution and the overall device usage time correlation of each of the health information feature vectors according to the health information feature matrix to obtain a single device usage time distribution feature sequence and an overall device usage time correlation feature sequence of the health information feature matrix; Based on the single device usage time distribution feature sequence of each smart wearable device and the overall device usage time association feature sequence, feature analysis of device usage trend, device usage periodicity, and device usage fluctuation is performed, and the analyzed device usage trend feature, device usage period feature, and device usage fluctuation feature are weighted fused to obtain the information collection mode of the health information feature matrix; Performing pattern matching processing on the information collection pattern according to several preset usage patterns stored in the reserve database to obtain matching indexes between the information collection pattern and various preset usage patterns; According to the matching index between the information collection mode and various preset usage modes, a mode weighted fusion process is performed on various preset usage modes in the reserve database to obtain the user's device usage mode for each smart wearable device.

5. The health management data fusion method based on multiple smart wearable devices as claimed in claim 1, characterized in that: The step of performing a health status trend analysis on the health information feature matrix at a continuous time level according to the device usage mode to obtain a health trend fitting feature distribution of the user comprises: Analyzing the health display form of the user according to the device usage mode to obtain the state display standards of several possible health states of the user corresponding to the device usage mode at the continuous time level, combining the state display standards of various possible health states of the user corresponding to the device usage mode at the continuous time level to obtain the continuous time evaluation standard of the health information feature matrix corresponding to the device usage mode; According to various state display standards in the continuous time evaluation standard, the health information feature matrix is ​​respectively subjected to feature correlation fitting verification at the continuous time level to obtain a health state standard fitting curve corresponding to various state display standards of the health information feature matrix; Performing multi-level fitting feature extraction on the health status standard fitting curves corresponding to various health status display standards of the health information feature matrix to obtain a highly fitting part, a normal fitting part and an abnormal fitting part of each health status standard fitting curve; According to the distribution proportions of the highly fitted part, the normal fitted part and the abnormally fitted part of each of the health status standard fitting curves, the confidence level of each of the health status standards is evaluated to obtain the state confidence level of the possible health state corresponding to each of the health status standards; An overall fitting method is analyzed for each of the possible health states according to the state confidence level of each of the possible health states, and health trend tendencies of each of the possible health states are extracted and combined according to the overall fitting method to obtain a health trend fitting feature distribution of the user.

6. The health management data fusion method based on multiple smart wearable devices as claimed in claim 3, characterized in that: The step of performing a predictive analysis on the health information feature matrix based on the health trend fitting feature distribution to obtain a predictive information feature matrix includes: Selecting trend key nodes for the health information feature matrix based on the health trend fitting feature distribution, so as to select trend key nodes from each health information feature vector in the health information feature matrix, and performing node linking on each trend key node to obtain a prediction starting subject; Performing a trend development simulation for the prediction starting subject in a future time period according to the health trend fitting characteristic distribution, so as to obtain a simulation state of the prediction starting subject corresponding to the health trend fitting characteristic distribution; An information interaction relationship between the prediction starting subject and the health information feature matrix is ​​analyzed, and based on the information interaction relationship, an information expansion analysis is performed on the simulation state to obtain a prediction information feature matrix.

7. The health management data fusion method based on multiple smart wearable devices as claimed in claim 5, characterized in that: The steps of performing information verification processing on the health information feature matrix generated in the future time according to the predicted information feature matrix, and performing feedback correction on the health trend fitting feature distribution according to the verification result to obtain the health trend verification feature distribution include: After obtaining the health information feature matrix generated at a future time, extracting substantially abnormal features of the health information feature matrix according to the prediction information feature matrix to obtain a matrix deviation tendency distribution between the health information feature matrix and the prediction information feature matrix; According to each possible health state corresponding to the health trend fitting characteristic distribution and the overall fitting method, the matrix deviation tendency distribution is subjected to the health state possibility feedback analysis to obtain the state confidence adjustment index of each possible health state fed back by the matrix deviation tendency distribution; The overall fitting method of the health trend fitting feature distribution is corrected according to the state confidence adjustment index of each possible health state fed back by the matrix deviation tendency distribution, and the health trend tendency of each possible health state is extracted and combined according to the corrected overall fitting method to obtain the user's health trend verification feature distribution.

8. The health management data fusion method based on multiple smart wearable devices as claimed in claim 1, characterized in that: The steps of integrating the health management data, the health information feature matrix, the health trend fitting feature distribution, and the health trend verification feature distribution in multiple dimensions to obtain a health management map for providing feedback on the user's health data include: Performing data compression and storage processing on the health management data, the health information feature matrix, the health trend fitting feature distribution, and the health trend verification feature distribution to obtain a data record level; Continuously updating and recording the health trend fitting feature distribution and the health trend verification feature distribution to obtain the health trend verification feature distribution updated in each time period, and performing overall comprehensive analysis and health status change analysis on the health trend verification feature distribution updated in each time period to obtain the user's overall health assessment conclusion and health status change assessment conclusion, and integrating information of the overall health assessment conclusion and health status change assessment conclusion to obtain a fusion assessment level; Based on the overall health assessment conclusion and the health status change assessment conclusion, a user health status confirmation detection method is analyzed for each of the smart wearable devices to obtain a detection execution plan for each of the smart wearable devices, and a guidance interaction interface is constructed according to the detection execution plan; The data recording level, the fusion evaluation level and the guidance interaction interface are combined to obtain the health management map.

9. A health management data fusion system based on multiple smart wearable devices, characterized in that: Used to implement a health management data fusion method based on multiple smart wearable devices as described in any one of claims 1-8.

Images