Method and system for predicting health state based on daily health detection data change

Abstract

The invention relates to a method and system for predicting a health state based on daily health detection data change, and the method comprises the following steps: collecting or receiving daily health detection data of a user, and carrying out the preprocessing of the daily health detection data; a time sequence analysis method is adopted to extract change trend features, periodic features and fluctuation features; health state features are mined based on a machine learning algorithm; constructing a health state prediction model according to the data, and training and verifying the health state prediction model by using historical data; inputting health detection data collected in real time into the trained health state prediction model to obtain a health state prediction result of the user; according to the invention, real-time monitoring and accurate prediction of the health state of the human body can be realized, health hidden dangers can be found earlier, and powerful support is provided for prevention and early intervention of diseases; the health state of the human body can be evaluated more comprehensively, and the accuracy and reliability of health prediction are improved.

Description

Technical Field

[0001] This invention relates to the field of health management technology, and more specifically, to a method and system for predicting health status based on changes in daily health monitoring data. Background Technology

[0002] With the improvement of people's living standards and the enhancement of health awareness, people are paying more and more attention to their own health. Traditional health monitoring methods often rely on regular physical examinations, which have problems such as long intervals and inability to detect potential health problems in a timely manner. While some health monitoring devices on the market can collect some health data in real time, their ability to deeply analyze the data and predict health status is limited, and they cannot provide users with comprehensive and accurate health warnings. There is a need for a method and system that can overcome the above-mentioned shortcomings and predict health status based on changes in daily health monitoring data. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for predicting health status based on changes in daily health monitoring data, and to provide a system for predicting health status based on changes in daily health monitoring data, in order to address the above-mentioned deficiencies of the prior art.

[0004] The technical solution adopted by this invention to solve its technical problem is:

[0005] A method for predicting health status based on changes in daily health monitoring data is constructed, wherein the method includes the following steps:

[0006] Collect or receive users' daily health monitoring data and preprocess the daily health monitoring data;

[0007] Time series analysis was used to analyze the time series of preprocessed daily health monitoring data and extract trend characteristics, periodic characteristics, and fluctuation characteristics.

[0008] Based on machine learning algorithms, health status features related to health status are extracted from preprocessed daily health monitoring data;

[0009] A health status prediction model is constructed based on health status characteristics and extracted trend, periodic, and fluctuation characteristics. Historical data is used to train and validate the health status prediction model.

[0010] The real-time collected health monitoring data is input into the trained health status prediction model to obtain the user's health status prediction result.

[0011] The method for predicting health status based on changes in daily health monitoring data according to the present invention, wherein the construction of the health status prediction model includes:

[0012] Based on the characteristics of health status and the data characteristics of the extracted data, such as changing trends, periodicity, and fluctuations, a suitable network architecture is selected.

[0013] Historical data is used to estimate and set the initial parameters of the network architecture to obtain a time series model;

[0014] The performance of time series models is evaluated and their parameters are optimized through residual analysis, information criteria, and one or more error metrics in model prediction.

[0015] By using established time series models to predict future health data, prediction intervals and confidence intervals are generated, providing a basis for health status assessment.

[0016] The method for predicting health status based on changes in daily health monitoring data described in this invention uses a convolutional neural network or a long short-term memory network as the network architecture.

[0017] The method for predicting health status based on changes in daily health monitoring data described in this invention includes data preprocessing, which includes one or more of the following: data cleaning, missing value handling, and data smoothing.

[0018] The method for predicting health status based on changes in daily health monitoring data described in this invention uses a moving average method, an exponential smoothing method, or a linear regression method to extract the trend of data changes.

[0019] The method for predicting health status based on changes in daily health monitoring data described in this invention uses Fourier transform, wavelet transform, or autocorrelation function and partial autocorrelation function analysis methods to extract periodic features.

[0020] The method for predicting health status based on changes in daily health monitoring data described in this invention involves extracting fluctuation characteristics by calculating statistical quantities or fluctuation models of the data.

[0021] The method for predicting health status based on changes in daily health monitoring data according to the present invention further includes:

[0022] Set an early warning threshold. When the user's predicted health status exceeds the warning threshold, issue a health warning message and health management suggestions to the user.

[0023] The method for predicting health status based on changes in daily health monitoring data described in this invention includes one or more of the following: body temperature, blood pressure, blood lipids, heart rate, and sleep status.

[0024] A system for predicting health status based on changes in daily health monitoring data is provided to implement the method described above for predicting health status based on changes in daily health monitoring data. The system includes a data acquisition unit, a data processing unit, a model building unit, and a health status prediction unit.

[0025] The data acquisition unit is used to collect or receive users' daily health monitoring data;

[0026] The data processing unit is used to preprocess daily health monitoring data; to analyze the time series of the preprocessed daily health monitoring data using time series analysis methods to extract trend features, periodic features and fluctuation features; and to mine health status features related to health status from the preprocessed daily health monitoring data based on machine learning algorithms.

[0027] The modeling unit is used to construct a health status prediction model based on health status characteristics and extracted trend characteristics, periodic characteristics and fluctuation characteristics, and to train and validate the health status prediction model using historical data.

[0028] The health status prediction unit is used to receive real-time collected health monitoring data, input it into the trained health status prediction model, and obtain the user's health status prediction result.

[0029] The beneficial effects of this invention are as follows:

[0030] 1. This invention enables real-time monitoring and accurate prediction of human health status. Compared with traditional health monitoring methods, it can detect potential health risks earlier, providing strong support for disease prevention and early intervention.

[0031] 2. By comprehensively analyzing the changing trends of various health monitoring data, this invention can more comprehensively assess the health status of the human body, avoid the limitations of monitoring a single indicator, and improve the accuracy and reliability of health prediction.

[0032] 3. Predictive models built on machine learning and deep learning algorithms have powerful data processing and analysis capabilities, enabling them to automatically learn and adapt to the health data characteristics of different users, and provide users with personalized health prediction and management solutions.

[0033] 4. The method of the present invention is simple and easy to implement, and can be widely applied in fields such as family health management, community healthcare, and telemedicine, and has significant social and economic value. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort:

[0035] Figure 1 This is a flowchart of a preferred embodiment of the method for predicting health status based on changes in daily health monitoring data according to the present invention.

[0036] Figure 2 This is a system principle block diagram of a preferred embodiment of the present invention for predicting health status based on changes in daily health monitoring data. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, a clear and complete description will be provided below in conjunction with the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0038] A preferred embodiment of the present invention provides a method for predicting health status based on changes in daily health monitoring data, such as... Figure 1 As shown, it includes the following steps:

[0039] S01: Collect or receive users' daily health monitoring data and preprocess the daily health monitoring data;

[0040] S02: Time series analysis is used to analyze the time series of preprocessed daily health monitoring data and extract trend characteristics, periodic characteristics and fluctuation characteristics;

[0041] S03: Based on machine learning algorithms, extract health status features related to health status from preprocessed daily health monitoring data;

[0042] S04: Construct a health status prediction model based on health status characteristics and extracted trend, periodic and fluctuation characteristics, and use historical data to train and validate the health status prediction model;

[0043] S05: Input the real-time collected health monitoring data into the trained health status prediction model to obtain the user's health status prediction result.

[0044] The beneficial effects of this invention are as follows:

[0045] 1. This invention enables real-time monitoring and accurate prediction of human health status. Compared with traditional health monitoring methods, it can detect potential health risks earlier, providing strong support for disease prevention and early intervention.

[0046] 2. By comprehensively analyzing the changing trends of various health monitoring data, this invention can more comprehensively assess the health status of the human body, avoid the limitations of monitoring a single indicator, and improve the accuracy and reliability of health prediction.

[0047] 3. Predictive models built on machine learning and deep learning algorithms have powerful data processing and analysis capabilities, enabling them to automatically learn and adapt to the health data characteristics of different users, and provide users with personalized health prediction and management solutions.

[0048] 4. The method of the present invention is simple and easy to implement, and can be widely applied in fields such as family health management, community healthcare, and telemedicine, and has significant social and economic value.

[0049] It should be noted that the order of steps S02 and S03 is not important, and they can be done in parallel.

[0050] This invention aims to address the problems existing in current health monitoring technologies and provides a method for predicting human health status based on changes in daily health monitoring data. By comprehensively analyzing the changing trends of multi-dimensional data such as body temperature, blood pressure, blood lipids, heart rate, and sleep status, it achieves accurate prediction of human health status, providing users with timely health warnings and personalized health management suggestions. Specific implementation details are as follows:

[0051] Data collection:

[0052] Smart wearable devices (such as smart bracelets and smartwatches) and home medical monitoring devices (such as electronic blood pressure monitors and blood lipid meters) can be used to collect users' daily health monitoring data, including data on body temperature, blood pressure, blood lipids, heart rate, and sleep status. The collected data should be characterized by high precision and high frequency to ensure the accuracy and completeness of the data.

[0053] Example Explanation: Taking a smart bracelet as an example, it is set to collect data every 10 minutes, including the user's body temperature, heart rate, and sleep status. Simultaneously, the user uses a home electronic blood pressure monitor to measure blood pressure three times daily and a blood lipid meter to measure blood lipids once a week.

[0054] Data preprocessing:

[0055] The collected data is cleaned and filtered to remove outliers and noisy data. For example, for blood pressure data, obvious errors caused by improper equipment operation or external interference are removed. This process is used to standardize the data, converting data of different dimensions and ranges to a uniform scale for easier subsequent analysis and processing. It also includes interpolating or estimating missing data points, using linear interpolation, polynomial interpolation, or filling in missing data points based on the average of adjacent data points. Moving averages or exponential smoothing methods can be used to smooth the data, reducing the interference of short-term fluctuations and making data trends more apparent.

[0056] Example Explanation: Outlier detection is performed on the collected data. For example, for body temperature data, if a data point's value is below 35℃ or above 42℃, it is identified as an outlier and removed. Then, the data is standardized, converting body temperature data into values ​​between 0 and 1, and blood pressure data into relative values ​​in standard blood pressure units, etc.

[0057] Data analysis and feature extraction:

[0058] By employing time series analysis, the time series of various health monitoring data are analyzed to extract the data's changing trends, periodic characteristics, and fluctuation characteristics. This method can capture the patterns of health data changes over time and promptly detect abnormal changes.

[0059] For example, analyzing the diurnal variation patterns of body temperature data and the long-term fluctuation trends of blood pressure data can provide early warnings of the occurrence of chronic diseases such as hypertension.

[0060] Trend Analysis:

[0061] 1. Moving Average Method: Calculates the average value of data over a certain time window to identify long-term trends. For example, a 7-day moving average can be used to analyze the weekly trend of blood pressure data.

[0062] 2. Linear Regression: Using time as the independent variable and health indicator data as the dependent variable, a straight line is fitted to analyze the linear trend of the data.

[0063] 3. Trend Separation: Use filters (such as Hodrick-Prescott filters) to separate trend components from the raw data in order to observe and analyze trends more clearly.

[0064] Periodic analysis:

[0065] 1. Fourier Transform: Converts time series data from the time domain to the frequency domain to identify periodic components in the data. By analyzing the spectrogram, the length of the main period is determined.

[0066] 2. Wavelet Transform: Performs multi-resolution analysis on data to identify periodic features at different time scales. Wavelet transform can simultaneously capture both local and global periodic changes in data.

[0067] 3. Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) Analysis: Calculate the autocorrelation and partial autocorrelation of the data to identify periodic patterns. For example, observe significant peaks in the ACF plot to determine periodic intervals.

[0068] Fluctuation characteristic analysis:

[0069] 1. Calculate statistics: Calculate statistics such as standard deviation, variance, and mean square deviation of data to quantify the degree of data fluctuation.

[0070] 2. Fluctuation Model Construction: Use autoregressive conditional heteroscedasticity (ARCH) models or generalized autoregressive conditional heteroscedasticity (GARCH) models to describe the fluctuation characteristics of the data, especially for data with conditional heteroscedasticity.

[0071] 3. Fluctuation feature extraction: Identify the amplitude, frequency, and duration of fluctuations, and analyze the regularity and potential influencing factors of fluctuations.

[0072] Based on machine learning algorithms, features related to health status are extracted from data. For example, cluster analysis can be used to divide users into different health risk groups, and association rule mining can be used to find potential correlations between different health indicators.

[0073] Example description: Time series analysis of body temperature data to extract its diurnal variation periodicity and short-term fluctuation characteristics; long-term trend analysis of blood pressure data to extract its upward or downward trend characteristics; seasonal variation analysis of blood lipid data, etc. Simultaneously, machine learning algorithms are used to uncover the correlation characteristics between different health indicators, such as the correlation between heart rate and blood pressure.

[0074] Health status prediction model construction:

[0075] Based on the extracted features and analysis results, a health status prediction model is constructed. This model can employ deep learning algorithms, such as Convolutional Neural Networks (CNNs) or Recurrent Neural Networks (RNNs) and their variants, Long Short-Term Memory Networks (LSTMs), to fully utilize the time-series characteristics and spatial correlations of the data. These algorithms possess powerful data processing and analysis capabilities, enabling them to automatically learn and adapt to the health data characteristics of different users, thereby improving prediction accuracy and generalization ability. Compared to traditional statistical analysis methods, deep learning algorithms can better handle complex data relationships and nonlinear features. Using a large amount of historical health data for model training and validation allows for optimization of model parameters and structure, further enhancing prediction accuracy and generalization ability.

[0076] Example Description: An LSTM network is chosen to build a health status prediction model. Preprocessed data is used as input to the model, and the user's health status classification (e.g., healthy, sub-healthy, pre-disease) is used as the model's output. Historical health data is used to train the model, and by adjusting the model's hyperparameters and optimizing the algorithm, the model's prediction accuracy is improved to over 90%.

[0077] The specific steps can be as follows:

[0078] (I) Data Preparation

[0079] Data collection and organization: Collect a large amount of historical health monitoring data, including multi-dimensional data such as body temperature, blood pressure, blood lipids, heart rate and sleep status, and label them (e.g., healthy, sub-healthy, disease status).

[0080] Data normalization: Normalize data of different dimensions and ranges so that the data fall within the same numerical range (such as between 0 and 1), thereby improving the stability and convergence speed of model training.

[0081] Dataset partitioning: Divide the dataset into training, validation, and test sets, typically in proportions of 70%, 15%, and 15%, to ensure that the training, validation, and testing processes of the model are independent of each other.

[0082] (II) Model Construction

[0083] Choosing a Deep Learning Architecture: Select a suitable deep learning model architecture based on the task requirements, such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), or Convolutional Neural Network (CNN). For time series data, LSTM and GRU are generally more suitable because they can effectively handle long-term dependencies in sequential data.

[0084] Model structure: Determine the structure of the input layer, hidden layer, and output layer. For example, the number of neurons in the input layer matches the number of input features; the hidden layer can contain multiple LSTM or GRU units to extract complex features from the data; the number of neurons in the output layer is determined according to the type of prediction task (such as the number of categories in a classification task or the number of predicted values ​​in a regression task).

[0085] Define the loss function and optimizer: Select appropriate loss functions (such as mean squared error, cross-entropy loss, etc.) and optimization algorithms (such as stochastic gradient descent, Adam optimizer, etc.) for model training and parameter updates.

[0086] (III) Model Training

[0087] Initialize model parameters: Initialize the model's weights and biases, usually using random initialization methods such as He initialization or Xavier initialization to accelerate model convergence.

[0088] Forward propagation: The training data is input into the model, and forward propagation calculations are performed through each layer of the network to obtain the model's predicted output.

[0089] Loss calculation: Based on the model's predicted output and the true labels, calculate the value of the loss function to evaluate the model's performance.

[0090] Backpropagation and parameter update: The gradient of the loss function with respect to the model parameters is calculated using the backpropagation algorithm, and the model parameters are updated using the optimizer to minimize the value of the loss function. This process is repeated on the training data for multiple epochs until the model converges.

[0091] (iv) Model Validation and Optimization

[0092] Validate model performance: Evaluate the model's performance on the validation set by calculating metrics such as accuracy, recall, and F1 score (for classification tasks) or mean squared error and mean absolute error (for regression tasks) to determine the model's ability to generalize on unseen data.

[0093] Adjusting hyperparameters: Based on the validation results, adjust the model's hyperparameters, such as learning rate, hidden layer size, and regularization parameters, to optimize model performance. Methods such as grid search, random search, or Bayesian optimization can be used for hyperparameter tuning.

[0094] Model optimization: Apply regularization techniques (such as L1 / L2 regularization, Dropout) and early stopping strategies to prevent model overfitting and improve the model's generalization ability.

[0095] (V) Model Deployment and Prediction

[0096] Model saving and loading: After model training is complete, the model's parameters and structure are saved to a file for later loading and use. In practical applications, the saved model is loaded into the prediction system.

[0097] Real-time data input: The real-time collected health monitoring data is preprocessed (e.g., normalized) and then input into the trained deep learning model.

[0098] Health status prediction: The model performs forward propagation calculations based on the input real-time data and outputs the user's health status prediction results, such as health status classification or specific health risk indicators.

[0099] Results Interpretation and Feedback: Interpret the prediction results, provide users with easy-to-understand health assessment reports, and offer more personalized health management advice based on the prediction results.

[0100] Health Status Prediction and Early Warning:

[0101] Real-time collected health monitoring data is input into a trained prediction model to obtain a prediction of the user's health status. The prediction results can include a classification of health status (e.g., healthy, sub-healthy, pre-disease), and specific health risk indicators; it can adapt to the health data characteristics of different users, providing personalized health predictions. For example, for users of different ages and lifestyles, the model can make accurate predictions based on their individual characteristics, providing more targeted health management suggestions.

[0102] Based on the prediction results, an early warning threshold is set. When a user's health status reaches or exceeds the warning threshold, a health warning message is promptly issued to the user, along with corresponding health management suggestions, such as lifestyle adjustments and medical advice. This real-time warning and personalized advice can help users take timely measures to prevent the occurrence and development of diseases, and improve the efficiency and effectiveness of health management.

[0103] Example Explanation: Real-time collected health monitoring data is input into a trained LSTM model to obtain a prediction of the user's health status. For example, if the model predicts that the user is in a sub-healthy state and that heart rate and blood pressure indicators show an upward trend, a health warning message is issued to the user, suggesting that the user pay attention to rest, adjust their diet, and have regular check-ups at the hospital.

[0104] This invention not only collects common health data such as body temperature, blood pressure, and heart rate, but also covers more comprehensive health indicators such as blood lipids and sleep status. This multi-dimensional data collection method can more comprehensively reflect the body's health status, avoiding the limitations of monitoring a single indicator. For example, blood lipid data can reflect the body's metabolic status, while sleep status data is closely related to the body's recovery and mental health. The comprehensive analysis of these data can more accurately assess health status. Through machine learning algorithms, this invention can uncover potential correlations between different health indicators. For example, it discovers the correlation between heart rate and blood pressure, and the link between blood lipid levels and sleep quality. This correlation analysis helps to understand the interactions between health indicators more deeply, providing richer evidence for health prediction.

[0105] Application prospects:

[0106] 1. Family Health Management: The method of this invention is simple and easy to implement, and can be widely applied in the field of family health management. Through smart wearable devices and home medical testing equipment, users can conveniently conduct daily health monitoring at home and obtain timely health warnings and management suggestions, which helps to improve the health awareness and health level of family members.

[0107] 2. Community Healthcare and Telemedicine: This invention is also applicable to the fields of community healthcare and telemedicine. Community doctors and telemedicine platforms can utilize the methods of this invention to monitor and analyze patients' health data in real time, promptly identify patients' health problems, and provide personalized medical services, thereby improving the efficiency and quality of medical services.

[0108] A system for predicting health status based on changes in daily health monitoring data is provided to implement the method described above for predicting health status based on changes in daily health monitoring data. Figure 2 As shown, the system includes a data acquisition unit 10, a data processing unit 11, a model building unit 12, and a health status prediction unit 13;

[0109] Data acquisition unit 10 collects or receives users' daily health monitoring data;

[0110] The data processing unit 11 is used to preprocess the daily health monitoring data; it uses time series analysis to analyze the time series of the preprocessed daily health monitoring data and extracts trend features, periodic features and fluctuation features; and it uses machine learning algorithms to mine health status features related to health status from the preprocessed daily health monitoring data.

[0111] Modeling unit 12 constructs a health status prediction model based on health status characteristics and extracted trend, periodic and fluctuation characteristics, and uses historical data to train and validate the health status prediction model.

[0112] The health status prediction unit 13 receives real-time collected health monitoring data, inputs it into the trained health status prediction model, and obtains the user's health status prediction result.

[0113] This invention enables real-time monitoring and accurate prediction of human health status. Compared with traditional health monitoring methods, it can detect potential health risks earlier, providing strong support for disease prevention and early intervention.

[0114] By comprehensively analyzing the changing trends of various health monitoring data, this invention can more comprehensively assess the health status of the human body, avoid the limitations of monitoring a single indicator, and improve the accuracy and reliability of health prediction.

[0115] Predictive models built on machine learning and deep learning algorithms have powerful data processing and analysis capabilities, enabling them to automatically learn and adapt to the health data characteristics of different users, and provide users with personalized health prediction and management solutions.

[0116] The system of this invention is simple and easy to implement, and can be widely used in fields such as family health management, community healthcare, and telemedicine, and has significant social and economic value.

[0117] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for predicting health status based on changes in daily health monitoring data, characterized in that, The method includes the following steps: Collect or receive users' daily health monitoring data and preprocess the daily health monitoring data; Time series analysis was used to analyze the time series of preprocessed daily health monitoring data and extract trend characteristics, periodic characteristics, and fluctuation characteristics. Based on machine learning algorithms, health status features related to health status are extracted from preprocessed daily health monitoring data; A health status prediction model is constructed based on health status characteristics and extracted trend, periodic, and fluctuation characteristics. Historical data is used to train and validate the health status prediction model. The real-time collected health monitoring data is input into the trained health status prediction model to obtain the user's health status prediction result.

2. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, The construction of the health status prediction model includes: Based on the characteristics of health status and the data characteristics of the extracted data, such as changing trends, periodicity, and fluctuations, a suitable network architecture is selected. Historical data is used to estimate and set the initial parameters of the network architecture to obtain a time series model; The performance of time series models is evaluated and their parameters are optimized through residual analysis, information criteria, and one or more error metrics in model prediction. By using established time series models to predict future health data, prediction intervals and confidence intervals are generated, providing a basis for health status assessment.

3. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, The network architecture employs either a convolutional neural network or a long short-term memory network.

4. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, Data preprocessing includes one or more of the following: data cleaning, missing value handling, and data smoothing.

5. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, The trend of data change can be extracted using the moving average method, exponential smoothing method, or linear regression method.

6. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, Periodic features are extracted using Fourier transform, wavelet transform, or autocorrelation and partial autocorrelation function analysis methods.

7. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, The extraction of fluctuation characteristics is achieved by calculating statistical measures of the data or using fluctuation models.

8. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, The method further includes: Set an early warning threshold. When the user's predicted health status exceeds the warning threshold, issue a health warning message and health management suggestions to the user.

9. The method for predicting health status based on changes in daily health monitoring data according to claim 1, characterized in that, The routine health monitoring data includes one or more of the following: body temperature, blood pressure, blood lipids, heart rate, and sleep status.

10. A system for predicting health status based on changes in daily health monitoring data, used to implement the method for predicting health status based on changes in daily health monitoring data as described in any one of claims 1-9, characterized in that, The system includes a data acquisition unit, a data processing unit, a model building unit, and a health status prediction unit; The data acquisition unit is used to collect or receive users' daily health monitoring data; The data processing unit is used to preprocess daily health monitoring data; This method is used to analyze the time series of preprocessed daily health monitoring data using time series analysis methods, and to extract trend characteristics, periodic characteristics, and fluctuation characteristics. It is also used to mine health status features related to health status from preprocessed daily health monitoring data based on machine learning algorithms; The modeling unit is used to construct a health status prediction model based on health status characteristics and extracted trend characteristics, periodic characteristics and fluctuation characteristics, and to train and validate the health status prediction model using historical data. The health status prediction unit is used to receive real-time collected health monitoring data, input it into the trained health status prediction model, and obtain the user's health status prediction result.

Images