User monitoring data analysis and early warning method based on real-time information synchronization
By using self-organizing monitoring networks and data analysis from wearable devices, the problem of high false alarm rates caused by individual differences has been solved, achieving real-time accuracy and personalized early warning for health monitoring.
Patent Information
- Application Number
- CN202411018849.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-07-29
AI Technical Summary
Current health monitoring technologies fail to take into account individual differences and cannot effectively distinguish between normal physiological fluctuations and abnormal health signals, resulting in a high false alarm rate.
By connecting grassroots stations to obtain single-disease health data, configuring a self-organizing monitoring network, using wearable devices to collect personal health data, conducting flexible and continuous analysis, predicting random and regular patterns, optimizing early warning false alarms, and iteratively updating and synchronously monitoring based on the analysis results.
It enables real-time data capture and analysis based on users' health history and characteristics, reducing false alarms and ensuring the real-time accuracy and effectiveness of health monitoring.
Smart Images

Figure CN118553445B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field related to data processing, and in particular to a user monitoring data analysis and early warning method based on real-time information synchronization. Background Art
[0002] In the field of health monitoring, conventional monitoring methods often have certain lags and limitations. For example, previous health monitoring relied on regular physical examinations or manually entered data, which could not reflect changes in the user's health status in real time and was slow to respond to sudden health events. With the development of the Internet of Things, big data and artificial intelligence technologies, wearable monitoring devices can be analyzed in real time to provide users with personalized health warnings.
[0003] However, real-time health monitoring also faces some technical challenges. For example, most wearable monitoring devices use a fixed threshold warning method without fully considering individual differences. In addition, how to accurately extract meaningful information from noisy data and how to effectively distinguish normal physiological fluctuations from abnormal health signals are key issues that need to be addressed.
[0004] In summary, the existing technology fails to take into account the individual differences between different users, cannot effectively distinguish normal physiological fluctuations from abnormal health signals, and has technical problems such as a high false alarm rate. Summary of the invention
[0005] This application provides a user monitoring data analysis and early warning method based on real-time information synchronization, aiming to solve the technical problems in the prior art that individual differences between different users cannot be taken into account, normal physiological fluctuations cannot be effectively distinguished from abnormal health signals, and there is a high false alarm rate.
[0006] In view of the above problems, the technical solution to implement this application is:
[0007] The present application provides a user monitoring data analysis and early warning method based on real-time information synchronization, wherein the method comprises: connecting to a grassroots site to obtain a target user's single disease health data, wherein the single disease health data comprises personal basic information, physiological indicators, and medical records; configuring a self-organizing monitoring network based on the single disease health data, wherein the self-organizing monitoring network automatically establishes connections between monitoring nodes according to the health status of the target user; based on the self-organizing monitoring network, using a wearable monitoring device to collect personal health data, wherein the personal health data comprises biometric data and user activity data; in the self-organizing monitoring network, through the personal health The health data is continuously analyzed for flexibility to predict the first sliding window and the first single disease false positive signal required for the random dominant mode and the regular dominant mode; in the self-organizing monitoring network, the health data of the single disease is continuously analyzed for flexibility to predict the second sliding window and the second single disease false positive signal required for the random dominant mode and the regular dominant mode; based on the first sliding window and the first single disease false positive signal and / or the second sliding window and the second single disease false positive signal, the health data of each single disease specified by the target user is optimized for early warning false alarms, iteratively updated with the goal of minimizing the health risk of the user, and synchronous monitoring and early warning are performed.
[0008] In summary, one or more technical solutions provided in this application connect to grassroots sites, obtain health data of single diseases, and configure a self-organizing monitoring network for automatic monitoring; use wearable devices to collect personal health data, analyze and predict random and regular patterns, and optimize early warning false alarms; based on the analysis results, optimize early warning false alarms for each single disease health data limited by the target user, iterate and update with the goal of minimizing user health risks, and monitor and warn synchronously. This solves the technical problems of failing to take into account individual differences between different users, failing to effectively distinguish between normal physiological fluctuations and abnormal health signals, and having a high false alarm rate. It realizes the real-time capture and analysis of user health data based on each user's health history and personal characteristics, establishes a self-organizing monitoring network, more accurately identifies health patterns and predicts anomalies, reduces false alarms, and ensures the technical effect of real-time and accurate monitoring and early warning of each user's health status. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Figure 1 A flowchart of a user monitoring data analysis and early warning method based on real-time information synchronization is provided for this application;
[0010] Figure 2 A flowchart of initialization configuration in a user monitoring data analysis and early warning method based on real-time information synchronization is provided for this application. DETAILED DESCRIPTION
[0011] The present application is described in detail below with reference to the accompanying drawings. Figure 1 As shown, the present application provides a user monitoring data analysis and early warning method based on real-time information synchronization, wherein the method includes:
[0012] S1: Connect to the grassroots sites to obtain the single disease health data of the target user, which includes basic personal information, physiological indicators, and diagnosis and treatment records.
[0013] Establish communication links with primary medical service sites, which are usually community health service centers or family doctor service points; collect single disease health records of target users from the sites, which usually include basic personal information (such as age, gender), physiological indicators (such as blood pressure, blood sugar level) and medical records (past medical history, treatment plan).
[0014] S2: Based on the single disease health data, a self-organizing monitoring network is configured, and the self-organizing monitoring network automatically establishes connections between monitoring nodes according to the health status of the target user.
[0015] According to the specific disease type of the target user (such as hypertension, diabetes), corresponding monitoring rules and parameters are set; the self-organizing monitoring network will automatically adjust the connection between monitoring nodes according to the user's real-time health status, such as increasing the frequency of blood sugar monitoring before and after meals to capture post-meal blood sugar fluctuations; the flexibility of the self-organizing monitoring network allows it to automatically expand or shrink when necessary, ensuring the comprehensiveness and effectiveness of monitoring coverage.
[0016] S3: Based on the self-organizing monitoring network, a wearable monitoring device is used to collect personal health data, where the personal health data includes biometric data and user activity data.
[0017] Users wear smart watches, health bracelets or other wearable devices that can continuously record biometric data (heart rate, blood pressure, blood oxygen saturation) and user activity data (number of steps, exercise type, sleep quality); via Bluetooth or Wi-Fi, these devices transmit data to the self-organizing monitoring network in real time to ensure the continuity and immediacy of the data.
[0018] S4: In the self-organizing monitoring network, the flexibility is continuously analyzed through the personal health data to predict the first sliding window and the first single disease false positive signal required for the random dominant mode and the regular dominant mode; S5: In the self-organizing monitoring network, the flexibility is continuously analyzed through the single disease health data to predict the second sliding window and the second single disease false positive signal required for the random dominant mode and the regular dominant mode; S6: Based on the first sliding window and the first single disease false positive signal and / or the second sliding window and the second single disease false positive signal, the early warning false alarm optimization is performed on the health data of each single disease specified by the target user, and iterative updates are performed with the goal of minimizing the user's health risks, and synchronous monitoring and early warning are performed.
[0019] Based on personal health data, continuous flexibility analysis is used to predict the first sliding window size required under the random dominant mode and the regular dominant mode, as well as the probability of occurrence of false positive signals of the first single disease. Similarly, continuous flexibility analysis is performed based on single disease health data to predict the second sliding window size and the probability of false positive signals of the second single disease.
[0020] Combine the false positive signal of the first sliding window and the first single disease with the false positive signal of the second sliding window and the second single disease to form a complete set of early warning signals; with the goal of minimizing user health risks, continuously optimize early warning signals through machine learning and data analysis, reduce false alarms, and improve the accuracy of early warnings; update early warning strategies in real time in the monitoring network to ensure that all nodes can receive the latest early warning information and achieve synchronous monitoring and early warning.
[0021] Furthermore, if Figure 2 As shown, the first sliding window and the first single disease false positive signal required for predicting the random dominant mode and the regular dominant mode, before that, the method of the present application also includes:
[0022] A time series analysis algorithm is used to predict short-term trends and initialize the regular dominant mode. Based on an autoregressive integrated moving average model, potential health status changes are identified and the random dominant mode is initialized.
[0023] A time series analysis algorithm is used to predict the short-term health trends of target users. Furthermore, user activity data is used to identify cyclic fluctuation patterns and predict the periodic changes of personal health data within the sliding window for subsequent adaptive optimization. This provides initial settings for regular governing patterns and helps understand possible periodic patterns in user health data.
[0024] Through the autoregressive integrated moving average model (ARIMA), potential changes in users' health status can be identified. Furthermore, the ARIMA model can identify non-periodic trends in health data, helping the system understand the impact of random events on users' health, thereby initializing random dominant patterns and helping to capture irregular or sudden changes in health status.
[0025] Furthermore, the regular control mode is initialized and configured, and the method of the present application includes:
[0026] Based on the user activity data in the personal health data, a cyclic fluctuation pattern is set; based on the cyclic fluctuation pattern, the periodic changes of the personal health data in the first sliding window are predicted; based on the first single disease false positive signal, combined with the periodic changes of the personal health data in the first sliding window, adaptive optimization is performed.
[0027] Collect personal health data from target users. This data mainly comes from user activity data, including daily exercise volume, sleep patterns, eating habits, etc. These data can reflect the user's lifestyle and daily behavior patterns; next, analyze the above user activity data to identify the cyclical fluctuation patterns therein, which involves using statistical methods or machine learning algorithms, such as time series analysis, to identify recurring periodic patterns in the data, such as weekly or monthly changes in exercise volume.
[0028] Predict the periodic changes of personal health data within the first sliding window. The sliding window is a data analysis technique that allows you to observe data patterns over time by sliding a window of fixed length across a data set. Predicting periodic changes means estimating how personal health data will change over time in a certain period of time in the future, which is usually based on the periodic patterns of historical data.
[0029] While predicting cyclical changes, it is also necessary to consider the false positive signals of the first disease type. A false positive signal refers to a situation where the monitoring system incorrectly reports the existence of a certain health problem. Analyzing these signals helps to understand under what circumstances the system may produce false alarms and the relationship between these false alarms and cyclical changes in personal health data.
[0030] Adaptively optimize the regular governing pattern. Further, adjust the monitoring strategy according to the false positive signals of the first single disease and the periodic changes of personal health data in the first sliding window. Specifically, the system needs to be able to distinguish between true health risk signals and false positive signals. Through continuous iteration, the system can more accurately identify health risks while reducing false alarms and ensuring that important health events are not missed.
[0031] Through the above steps, the regular control mode can be effectively initialized and configured, making the user monitoring data analysis and early warning method more accurate and reliable.
[0032] Furthermore, in combination with the periodic changes of the personal health data in the first sliding window, adaptive optimization is performed, and the method of the present application includes:
[0033] Analyze the periodic changes of personal health data within the first sliding window to match the normal fluctuation range and the abnormal signal threshold; based on the first single disease false positive signal, adjust the first sensitivity of the regular dominant pattern according to the normal fluctuation range; based on the first single disease false positive signal, adjust the second sensitivity of the regular dominant pattern according to the abnormal signal threshold; perform machine learning training on the regular dominant pattern through the first sensitivity and the second sensitivity.
[0034] For the personal health data collected within the first sliding window, in-depth analysis is conducted to identify the characteristics of periodic changes, which involves the application of time series analysis techniques, such as the autoregressive integrated moving average model (ARIMA), to detect and quantify regular fluctuations in the data. At the same time, a normal fluctuation range needs to be defined. This range is based on the statistical characteristics of historical data, such as the mean and standard deviation, to determine whether the data points are within the expected range of variation.
[0035] Based on the analysis of false positive signals of the first single disease, the first sensitivity of the regular governing pattern is adjusted. A false positive signal refers to a situation where the monitoring system mistakenly issues an alarm when there is actually no health risk. For example, if too many false positive signals are found, it indicates that the current sensitivity is too high, resulting in too many false alarms. Therefore, the first sensitivity needs to be appropriately lowered based on the normal fluctuation range to ensure that the system triggers an alarm only when it actually exceeds the expected range.
[0036] Similarly, the abnormal signal threshold is used to distinguish normal health data fluctuations from abnormal conditions that may indicate health risks. When the false positive signal of the first disease indicates frequent false alarms, the second sensitivity of the regular governing pattern should be adjusted according to the abnormal signal threshold. If the threshold is set too low, it may lead to too many false alarms, while if it is too high, it may lead to underreporting of actual health risks. Therefore, the second sensitivity is adjusted to find a balance point that can capture real health risks in a timely manner while avoiding excessive alarms.
[0037] The first sensitivity and the second sensitivity are used as input parameters for machine learning training of regular dominant patterns, which usually involves defining a loss function, such as mean square error (MSE), to measure the difference between the model prediction value and the actual health data; by minimizing the loss function, the sensitivity parameters are adjusted to optimize the performance of the regular dominant pattern; during the training process, the prediction function form of the Fourier series can be used to capture the periodic components in the data, and through optimization algorithms such as gradient descent, the model parameters are gradually adjusted until the optimal state is reached.
[0038] The entire process is an iterative cycle, starting with periodic analysis of data, adjusting sensitivity parameters to reduce false alarms, and finally optimizing regular patterns through machine learning training; each round of iteration is based on the latest understanding of personal health data, gradually improving the accuracy and reliability of the early warning system, and adaptive optimization ensures that the early warning mechanism can adjust itself as the individual's health status changes, providing the most effective monitoring and early warning services.
[0039] Furthermore, by using the first sensitivity and the second sensitivity, the regular governing pattern is subjected to machine learning training, and the method of the present application includes:
[0040] Definition of loss function in machine learning: , where N is the number of samples, is the actual value of the i-th sample, is the model prediction value; the first sensitivity is recorded as , the second sensitivity is recorded as , construct the objective function:
[0041] ;in, It depends on the first sensitivity. and the second sensitivity is denoted as The prediction function of
[0042] is the Fourier coefficient, ω is the fundamental frequency, and t is the time variable.
[0043] Define the loss function for machine learning, which is a criterion for evaluating the difference between the model's predicted values and the actual values. The expression of the loss function is: , where N is the number of samples, is the actual value of the i-th sample, For the model prediction value, the loss function is used to calculate the mean square error (MSE), which is a common indicator to measure the prediction accuracy of the model.
[0044] The first sensitivity is denoted as , the second sensitivity is recorded as , construct the objective function:
[0045] ; further, It depends on the first sensitivity. and the second sensitivity is denoted as The prediction function of .
[0046] Furthermore, ,in, , is the Fourier coefficient, ω is the fundamental frequency, and t is the time variable; the time series characteristics of health data are fitted by Fourier series, among which different first sensitivities and the second sensitivity This will affect the sensitivity of the model and thus the accuracy of the prediction.
[0047] During training, by adjusting the first sensitivity and the second sensitivity The value of , finding the combination that minimizes the loss function, usually involves using an optimization algorithm, such as gradient descent, to iteratively update the first sensitivity and the second sensitivity , until the objective function Reach the minimum.
[0048] The collected personal health data is used as a training set, and the model is trained using the objective function and prediction function constructed above. After the training is completed, an independent test set is used to verify the generalization ability of the model to ensure that the model not only performs well on the training data, but also can accurately predict new data that has not been seen. After the model is trained and verified, it is deployed in the actual environment to make real-time predictions of personal health data. According to the prediction results, the first sensitivity is continuously adjusted. and the second sensitivity The value of further optimizes the model performance and ensures the accuracy and response speed of the early warning system.
[0049] Through the above steps, the machine learning training of regular governing patterns in the user monitoring data analysis and early warning method based on real-time information synchronization is realized, ensuring that the system can effectively identify and predict abnormal changes in personal health data, issue early warnings in a timely manner, reduce false alarms, and improve the accuracy of early warnings.
[0050] Furthermore, the self-organizing monitoring network automatically establishes connections between monitoring nodes according to the health status of the target user. The method of the present application includes:
[0051] In the rule-governed mode, multiple first sliding windows are traversed to analyze the fluctuation characteristics of personal health data in different time periods, and the size of the first sliding window is M bytes; based on the fluctuation characteristics of personal health data in different time periods, the size of the first sliding window is dynamically matched with the goal of maximizing fitness, wherein the size of the first sliding window is controlled by adjusting the window size field in the TCP protocol; in the self-organizing monitoring network, the flexibility is continuously analyzed through the personal health data, and the self-organizing monitoring network is updated to reconfigure the connections between monitoring nodes.
[0052] A first sliding window with an initial size of M bytes is set to capture continuous segments of personal health data; in a regularity-dominated mode, the sliding window is used to sequentially scan different time periods of the personal health data; for each time period within the sliding window, the fluctuation pattern of the personal health data is analyzed, including identifying any periodic changes, abnormal conditions or trends.
[0053] Based on the analysis results, the adaptability of the current sliding window size to the data fluctuation characteristics is evaluated, with the goal of minimizing the false alarm rate while maintaining the sensitivity of the warning; using the window size field of the TCP protocol, the size of the first sliding window is dynamically adjusted. If the data changes rapidly, a smaller window is required to capture details; conversely, if the data is relatively stable, the window can be increased to reduce the computational burden; the above process is repeated until the window size that best suits the current data fluctuation characteristics is found.
[0054] In the self-organizing monitoring network, continuous flexibility analysis continues, that is, evaluating the network's response speed and accuracy to changes in personal health data under different window sizes; based on the analysis results, update the parameters of the self-organizing monitoring network, such as node weights or connection topology, to enhance the network's adaptability and early warning efficiency; based on the optimized network structure, reconfigure the connections between monitoring nodes to ensure efficient data transmission and real-time data processing.
[0055] Continuous monitoring and adjustment are repeated over and over again to respond to changes in personal health data over time, ensuring the long-term effectiveness and accuracy of the early warning system; the early warning results are compared with the actual health status to further optimize the sliding window size and network configuration; real-time analysis of individual health data and adaptive optimization in a dynamic environment are performed to improve the overall effectiveness and reliability of the early warning system; by fine-tuning the sliding window size and monitoring the network structure, the system can more accurately identify health risks, reduce unnecessary warnings, and ensure that important health events are responded to in a timely manner.
[0056] Furthermore, the iterative update is carried out with the goal of minimizing the health risks of users, and synchronous monitoring and early warning are carried out. The application method includes:
[0057] In the rule-governed mode, multiple false positive signals of the first single disease are traversed to extract correlation influence indicators of the user's health risk, wherein the correlation influence indicators include the false positive signal frequency, the false positive signal strength and the correlation with the positive signal of the first single disease, and the positive signal of the first single disease is associated with multiple false positive signals of the first single disease; based on the correlation influence indicators and in combination with multiple false positive signals of the first single disease, a multi-level early warning mechanism adapted to the target user is generated; based on the multi-level early warning mechanism, iterative updates are performed with the goal of minimizing the user's health risk.
[0058] In the regular dominance mode, it is necessary to traverse all false positive signals of the first disease type, which may be triggered by personal health data collected by wearable monitoring devices, and extract the following key correlation influencing indicators, including: false positive signal frequency (recording the number of false positives that occur per unit time, reflecting the frequency of false alarms of the monitoring system for non-real health problems), false positive signal intensity (quantifying the severity of each false positive alarm, which may be related to the degree of abnormality of the monitoring data), and correlation with the positive signal of the first disease type (analyzing the correlation between false positive signals and actual positive signals helps to understand the false alarm pattern of the monitoring system).
[0059] Based on the extracted correlation impact indicators and multiple first-type false positive signals of the disease, a multi-level early warning mechanism adapted to the target users is developed to distinguish different levels of health risks. Different warning levels are set according to signal strength and frequency, such as low, medium, and high risk levels. Furthermore, the warning threshold is dynamically adjusted according to the user's historical health data and behavior patterns to ensure that the early warning mechanism can accurately reflect changes in the user's health status.
[0060] Collect feedback from the actual operation of the early warning system, including false alarms, missed alarms and correct alarms; adjust the parameters of the early warning mechanism according to the feedback results, such as modifying thresholds, optimizing algorithms, etc., to reduce false alarms and missed alarms; use machine learning algorithms to enable the system to continuously learn and adapt to changes in the health patterns of target users, thereby improving the accuracy of early warnings.
[0061] Wearable monitoring devices continuously collect personal health data and upload it to the self-organizing monitoring network in real time; once the monitoring data triggers the early warning mechanism, an alert is immediately sent to the user so that timely action can be taken.
[0062] Overall, through continuous iteration and learning, the system can more accurately identify health risks, reduce unnecessary alarms, and ensure that users' health risks are minimized. It combines machine learning, network communications, and user interface design to ensure the efficiency and accuracy of the early warning system.
[0063] Furthermore, the random dominance mode is initialized and configured, and the method of the present application includes:
[0064] Based on the basic personal information in the single disease health data, identify the marked population, which includes pregnant women, chronic disease patients, and people with a genetic disease history; set a random fluctuation mode by comparing the identification information of the marked population and combining it with the user activity data in the personal health data.
[0065] Utilize basic personal information in single disease health data, which usually includes key data such as age, gender, and medical history; identify individuals belonging to the labeled population, including but not limited to pregnant women, chronic disease patients, and people with a history of genetic diseases; these classifications are defined based on medical knowledge and clinical experience, and involve risk factors, genetic tendencies, or physiological states for specific diseases.
[0066] By comparing personal basic information with the standard identification information of the marked population, it is confirmed whether the individual belongs to a certain marked population category; combined with user activity data in personal health data, including daily exercise, sleep patterns, eating habits, etc., to fully understand the individual's lifestyle and health behavior.
[0067] Based on the specific characteristics of the labeled population and user activity data, a random fluctuation pattern suitable for the population is developed. The random fluctuation pattern should take into account the individual's health status and lifestyle to reflect the natural variation range of their health data; adjust the parameters of the random fluctuation pattern to ensure that it can sensitively capture abnormal fluctuations in health data while reducing the occurrence of false positives.
[0068] The set random fluctuation pattern is applied to the real-time monitoring system as a benchmark for judging whether the health data deviates from the normal range; based on the subsequently collected data and system feedback, the parameters of the random fluctuation pattern are continuously fine-tuned to adapt to changes in individual health status and optimize system performance.
[0069] Regularly evaluate the effectiveness of the random fluctuation pattern and check its performance in identifying early health risks; based on the evaluation results, make necessary adjustments to the random fluctuation pattern and its initialization configuration process to improve the accuracy and reliability of the early warning system.
[0070] Overall, the initial configuration of the random dominance model is an iterative and dynamic process, which requires the combination of personal health data and user activity data, through careful analysis and timely adjustments, to ensure that the system can effectively monitor and warn of health risks of the marked population.
[0071] In summary, the beneficial effects of the embodiments of the present application are:
[0072] 1. Through real-time information synchronization, users’ health data can be monitored instantly and abnormal situations can be responded to quickly. By using advanced data analysis algorithms, health risks can be predicted more accurately, a self-organizing monitoring network can be established, and false alarms can be reduced.
[0073] 2. Based on the user's personal health data and historical records, more accurately identify health patterns and predict anomalies, generate customized warning thresholds, and improve the pertinence and effectiveness of warnings.
[0074] 3. The system can automatically adjust the early warning mechanism according to changes in the user's health patterns to minimize long-term health risks.
[0075] 4. Due to the use of the rule-governing mode, multiple false positive signals of the first single disease are traversed to extract the correlation influence index of the user's health risk, wherein the correlation influence index includes the false positive signal frequency, the false positive signal strength and the correlation with the positive signal of the first single disease, and the positive signal of the first single disease is associated with multiple false positive signals of the first single disease; based on the correlation influence index, a multi-level early warning mechanism adapted to the target user is generated in combination with multiple false positive signals of the first single disease; based on the multi-level early warning mechanism, iterative updates are performed with the goal of minimizing the health risk of the user. Through continuous iteration and learning, the system can more accurately identify health risks, reduce unnecessary alarms, and ensure the minimization of the health risk of the user, and combine machine learning, network communication and user interface design to ensure the efficiency and accuracy of the early warning system.
[0076] In summary, any step can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor, and no unnecessary restrictions are made here.
[0077] Furthermore, the above-mentioned technical scheme only reflects the preferred technical scheme of the technical scheme of the embodiment of the present application. Some changes that may be made to some parts thereof by technicians in this technical field all reflect the novel principles of the embodiment of the present application. Obviously, technicians in this field can make various changes and modifications to the present application without departing from the scope of the present application.
Claims
1. A user monitoring data analysis and early warning method based on real-time information synchronization, characterized in that: The method comprises: Connect to grassroots sites to obtain single disease health data of target users, including basic personal information, physiological indicators, and diagnosis and treatment records; Based on the single disease health data, a self-organizing monitoring network is configured, and the self-organizing monitoring network automatically establishes connections between monitoring nodes according to the health status of the target user; Based on the self-organizing monitoring network, using wearable monitoring devices to collect personal health data, the personal health data includes biometric data and user activity data; In the self-organizing monitoring network, the flexibility is continuously analyzed through the personal health data to predict the first sliding window required for the random dominant mode and the regular dominant mode and the first single disease false positive signal; In the self-organizing monitoring network, the single disease health data is used to continuously analyze the flexibility, and the second sliding window required for the random dominant mode and the regular dominant mode and the second single disease false positive signal are predicted; Based on the first sliding window and the first single disease false positive signal and / or the second sliding window and the second single disease false positive signal, the health data of each single disease defined by the target user is optimized for early warning false alarms, iteratively updated with the goal of minimizing the user's health risk, and synchronous monitoring and early warning are performed; Among them, the first sliding window and the first single disease false positive signal required for predicting the random dominant mode and the regular dominant mode also include: Using time series analysis algorithms to predict short-term trends and initialize configurations of the regular governing patterns; Based on an autoregressive integrated moving average model, potential health state changes are identified and the random dominant mode is initialized and configured; Initializing the configuration of the regular governing mode includes: Setting a cyclic fluctuation pattern based on user activity data in the personal health data; Based on the cyclic fluctuation pattern, predicting the periodic changes of the personal health data within the first sliding window; Based on the false positive signal of the first single disease, combined with the periodic changes of the personal health data in the first sliding window, adaptive optimization is performed; Combining the periodic changes of the personal health data in the first sliding window, adaptive optimization is performed, including: Analyze the periodic changes of the personal health data in the first sliding window to match the normal fluctuation range and abnormal signal threshold; Based on the first single disease false positive signal, adjusting the first sensitivity of the regular governing mode according to the normal fluctuation range; Based on the first single disease false positive signal, adjusting the second sensitivity of the regular dominant mode according to the abnormal signal threshold; Performing machine learning training on the regular governing pattern through the first sensitivity and the second sensitivity; The method of performing machine learning training on the regular governing pattern by using the first sensitivity and the second sensitivity includes: Definition of loss function in machine learning: , where N is the number of samples, is the actual value of the i-th sample, is the model prediction value; The first sensitivity is denoted as , the second sensitivity is recorded as , construct the objective function: ; in, It depends on the first sensitivity. and the second sensitivity is denoted as The prediction function of , , , is the Fourier coefficient, ω is the fundamental frequency, and t is the time variable; The self-organizing monitoring network automatically establishes connections between monitoring nodes according to the health status of the target user, including: In the rule-governing mode, multiple first sliding windows are traversed to analyze the fluctuation characteristics of personal health data in different time periods, and the size of the first sliding window is M bytes; Based on the fluctuation characteristics of personal health data in different time periods, dynamically matching the size of the first sliding window with the goal of maximizing fitness, wherein the size of the first sliding window is controlled by adjusting the window size field in the TCP protocol; In the self-organizing monitoring network, continuously analyzing the flexibility through the personal health data, updating the self-organizing monitoring network, and reconfiguring the connections between monitoring nodes; Iterate and update with the goal of minimizing user health risks, and conduct synchronous monitoring and early warning, including: In the rule-governing mode, multiple false positive signals of a first single disease are traversed to extract correlation impact indicators of user health risks, wherein the correlation impact indicators include false positive signal frequency, false positive signal strength, and correlation with the first single disease positive signal, and the first single disease positive signal is associated with multiple first single disease false positive signals; Based on the correlation impact indicator, combined with multiple first single disease false positive signals, a multi-level early warning mechanism adapted to the target user is generated; Based on a multi-level early warning mechanism, iterative updates are carried out with the goal of minimizing user health risks; Initializing and configuring the random dominance pattern includes: Based on the basic personal information in the single disease health data, identify the marked population, which includes pregnant women, chronic disease patients, and people with genetic disease history; The random fluctuation mode is set by comparing the identification information of the marked population and combining the user activity data in the personal health data.
Citation Information
Patent Citations
Big data-based health condition analysis and prediction method and system
CN111785374A
Method, apparatus and system for automatically processing medical images to output alarm
CN115861656A
Personalized nursing planning method and system for rehabilitation patients
CN117457147A
Respirator alarm management method and system for intensive care unit
CN117612725A
Night blood glucose prediction model construction method based on deep multi-task learning
CN117976229A
