A blood sugar risk warning method and system based on multiple parameters

Abstract

The present invention proposes a multi-parameter-based blood sugar risk warning method and system. The multi-parameter-based blood sugar risk warning method includes: retrieving the patient's medical data information from the medical terminal, and obtaining the first blood sugar data information and the second blood sugar data information from the medical data information; using the first risk prediction model and the second risk prediction model to perform data analysis on the first blood sugar data information and the second blood sugar data information, respectively, to obtain the corresponding first risk prediction value and the second risk prediction value; according to the first risk prediction value, the patient is predicted for the first blood sugar risk, and it is judged whether the patient meets the first blood sugar risk standard; when the patient does not meet the first blood sugar risk standard, the patient is predicted for the second blood sugar risk using the first risk prediction value and the second risk prediction value, and it is judged whether the patient meets the second blood sugar risk standard. The system includes modules corresponding to the steps of the method.

Description

Technical Field

[0001] The present invention proposes a multi-parameter-based blood sugar risk early warning method and system, belonging to the technical field of blood sugar risk prediction. Background Art

[0002] With the change of lifestyle in modern society, the incidence of metabolic diseases such as diabetes has increased year by year, and blood sugar management has become an important issue in public health. Traditional blood sugar monitoring methods mainly rely on a single blood sugar value to judge the patient's blood sugar status. However, this method often cannot fully reflect the patient's true blood sugar risk. In order to more accurately and comprehensively assess the patient's blood sugar risk, a new blood sugar risk warning method is needed.

[0003] In existing medical technologies, although there are some disease risk assessment methods based on multiple parameters, these methods often do not fully consider the multiple factors related to blood sugar, or simply linearly combine these factors, and fail to deeply explore the intrinsic connection and mutual influence between the factors. Therefore, the existing methods still need to be improved in improving the accuracy of blood sugar risk warning. Summary of the invention

[0004] The present invention provides a multi-parameter-based blood sugar risk warning method and system to solve the technical problems in the above-mentioned prior art. The technical solutions adopted are as follows:

[0005] A multi-parameter-based blood sugar risk early warning method, the multi-parameter-based blood sugar risk early warning method comprising:

[0006] Retrieving medical data information of the patient from the medical terminal, and acquiring first blood sugar data information and second blood sugar data information from the medical data information;

[0007] Using the first risk prediction model and the second risk prediction model to respectively analyze the first blood sugar data information and the second blood sugar data information, and obtain a first risk prediction value and a second risk prediction value corresponding to the first blood sugar data information and the second blood sugar data information;

[0008] Performing a first blood glucose risk prediction on the patient according to the first risk prediction value, and determining whether the patient meets the first blood glucose risk standard;

[0009] When the patient does not meet the first blood sugar risk standard, a second blood sugar risk prediction is performed on the patient using the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard.

[0010] Further, retrieving medical data information of the patient from the medical terminal, and acquiring first blood sugar data information and second blood sugar data information from the medical data information, including:

[0011] Retrieving the patient's medical data information from the medical terminal;

[0012] Extracting multi-parameter blood sugar data information from the medical data information, wherein the multi-parameter blood sugar data information includes fasting blood sugar value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value, and data collection time sequence corresponding to the fasting blood sugar value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value;

[0013] dividing the fasting blood glucose value in the multi-parameter blood glucose data information and its corresponding data collection time sequence into first blood glucose data information;

[0014] The glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value in the multi-parameter blood glucose data information, as well as the data collection timing corresponding to the glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value are divided into the second blood glucose data information.

[0015] Further, using the first risk prediction model and the second risk prediction model to respectively analyze the first blood glucose data information and the second blood glucose data information to obtain a first risk prediction value and a second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information, including:

[0016] Inputting the first blood glucose data information into a first risk prediction model for risk prediction to obtain a first risk prediction value; wherein the first risk prediction model adopts a convolutional neural network model, and the model structure of the first risk prediction model includes a data input end, a TCN feature extraction network, an XGBoost time series prediction layer and a data output end;

[0017] The second blood glucose data information is input into a second risk prediction model for risk prediction to obtain a second risk prediction value; the second risk prediction model adopts a binary logistic regression model.

[0018] Further, inputting the first blood glucose data information into a first risk prediction model for risk prediction to obtain a first risk prediction value includes:

[0019] inputting the first blood glucose data information into a first risk prediction model;

[0020] Obtaining a blood glucose curve graph of the patient's corresponding future days through the first risk prediction model; wherein the future days have a value range of 15-30 days;

[0021] The first risk prediction value is obtained according to the blood glucose curve corresponding to the patient in the future days combined with the blood glucose standard range value; wherein the first risk prediction value is obtained by the following formula:

[0022]

[0023] Where R represents the first risk prediction value; X i represents the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model; n represents the number of input feature variables corresponding to the TCN feature extraction network; X up and X down They respectively represent the upper and lower limits of the healthy standard range of fasting blood sugar values.

[0024] Further, inputting the second blood glucose data information into a second risk prediction model for risk prediction to obtain a second risk prediction value includes:

[0025] Retrieve the binary logistic regression model and the multi-factor logistic regression analysis table corresponding to glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value;

[0026] Extracting the regression coefficients corresponding to glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value from the multi-factor logistic regression analysis table;

[0027] The binary logistic regression model is combined with the regression coefficients corresponding to glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value to obtain a second risk prediction value, wherein the second risk prediction value is obtained by the following formula:

[0028]

[0029] Among them, P c represents the second risk prediction value; β0 represents the intercept term; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j A data value representing the j-th second blood sugar data information;

[0030] Further, performing a first blood glucose risk prediction on the patient according to the first risk prediction value to determine whether the patient meets the first blood glucose risk standard includes:

[0031] Comparing the first risk prediction value with a preset first risk threshold;

[0032] When the first risk prediction value exceeds a preset first risk threshold, it is determined that the patient has reached the first blood sugar risk standard, and a blood sugar risk warning is issued;

[0033] When the first risk prediction value does not exceed the preset first risk threshold, it is determined that the patient does not meet the first blood sugar risk standard.

[0034] Further, when the patient does not meet the first blood sugar risk standard, a second blood sugar risk prediction is performed on the patient using the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard, including:

[0035] When the patient does not meet the first blood sugar risk standard, a second risk prediction value is retrieved;

[0036] comparing the second risk prediction value with a preset second risk threshold;

[0037] When the second risk prediction value exceeds a preset second risk threshold, the first comprehensive risk model is used in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard;

[0038] When the second risk prediction value does not exceed the preset second risk threshold, the second comprehensive risk model is used to combine the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard.

[0039] Further, when the second risk prediction value exceeds a preset second risk threshold, the first comprehensive risk model is used in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard, including:

[0040] When the second risk prediction value exceeds the preset second risk threshold, the weight value is obtained by using the second risk prediction value and the second risk threshold, wherein the weight value is obtained by the following formula:

[0041]

[0042] Where w(R) represents the weight value; R represents the first risk prediction value; R y represents the first risk threshold; P represents the second risk prediction value; P y represents the second risk threshold;

[0043] The first comprehensive risk model is obtained by using the weight value; wherein the structure of the first comprehensive risk model is as follows:

[0044]

[0045] Among them, P01 represents the first comprehensive risk parameter obtained by the first comprehensive risk model; β0 represents the intercept term; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j A data value representing the j-th second blood sugar data information;

[0046] Substituting the first risk prediction value and the second blood glucose data information into a first comprehensive risk model to obtain a first comprehensive risk parameter;

[0047] comparing the first comprehensive risk parameter with a preset first comprehensive parameter threshold;

[0048] When the first comprehensive risk parameter exceeds the preset first comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard and a blood sugar risk warning is issued.

[0049] Further, when the second risk prediction value does not exceed the preset second risk threshold, the second comprehensive risk model is used to combine the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard, including:

[0050] When the second risk prediction value does not exceed the preset second risk threshold, calling the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network;

[0051] The first risk prediction value and the input feature variable corresponding to the TCN feature extraction network are used to obtain a complex evaluation function, wherein the structure of the complex evaluation function is as follows:

[0052]

[0053] Among them, f(R, X) represents the complex evaluation function; n represents the number of input feature variables corresponding to the TCN feature extraction network; X i represents the predicted fasting blood glucose value outputted by the i-th first risk prediction model; Y j represents the data value corresponding to the j-th second blood glucose data information; α i represents the weight coefficient between the predicted fasting blood glucose value outputted by the i-th first risk prediction model and the first risk prediction value R; θ ij represents the average value of the predicted fasting blood glucose value outputted by the i-th first risk prediction model and the j-th second blood glucose data information and the weight value of the first risk prediction value R; R represents the first risk prediction value; m represents the number of second blood glucose data information;

[0054] The influence association function is obtained by using the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network, wherein the structure of the influence association function is as follows:

[0055]

[0056] Among them, g(R, X) represents the influence correlation function; n represents the number of input feature variables corresponding to the TCN feature extraction network; X i represents the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model; R represents the first risk prediction value; δ represents the interaction coefficient between the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model and the first risk prediction value R, indicating the degree of influence of the characteristic variable and the first risk prediction value R on the second risk prediction value, and the interaction term reflects how the performance of a specific feature at different risk levels affects the risk prediction;

[0057] The complex evaluation function and the impact association function are used to obtain a second comprehensive risk model, wherein the structure of the second comprehensive risk model is as follows:

[0058]

[0059] Among them, P 02 represents the second comprehensive risk parameter obtained by the second comprehensive risk model; β 01 represents the intercept term, which means the value of logit(p) of the binary logistic regression model when the predicted fasting blood glucose value and the first risk prediction value of the output corresponding to the i-th first risk prediction model of all characteristic variables are both 0; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j represents the data value of the j-th second blood sugar data information; β R represents the regression coefficient of the first risk prediction value; γ represents the quadratic coefficient of the first risk prediction value, which is used to capture the nonlinear effect of the first risk prediction value R on logit(p). It represents the influence of the square of the first risk prediction value R on the final predicted risk, reflecting the impact of the nonlinear effect on the model when the risk value is high or low; g(R, X) represents the impact correlation function; f(R, X) represents the complex evaluation function; λ represents the adjustment coefficient corresponding to the complex evaluation function, and the range of the adjustment coefficient is [-1,1];

[0060] Substituting the first risk prediction value and the second blood glucose data information into a second comprehensive risk model to obtain a second comprehensive risk parameter;

[0061] comparing the second risk comprehensive parameter with a preset second comprehensive parameter threshold;

[0062] When the second risk comprehensive parameter exceeds the preset second comprehensive parameter threshold, it is determined that the patient reaches the second blood sugar risk standard and a blood sugar risk warning is issued.

[0063] A multi-parameter-based blood sugar risk warning system, the multi-parameter-based blood sugar risk warning system comprising:

[0064] a blood sugar data information acquisition module, used to retrieve the patient's medical data information from the medical terminal, and acquire first blood sugar data information and second blood sugar data information from the medical data information;

[0065] a risk prediction value acquisition module, used to perform data analysis on the first blood glucose data information and the second blood glucose data information respectively using the first risk prediction model and the second risk prediction model, and acquire the first risk prediction value and the second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information;

[0066] A first blood sugar risk standard determination module, used to perform a first blood sugar risk prediction for the patient according to the first risk prediction value, and determine whether the patient meets the first blood sugar risk standard;

[0067] The second blood sugar risk standard determination module is used to perform a second blood sugar risk prediction on the patient using the first risk prediction value and the second risk prediction value when the patient does not meet the first blood sugar risk standard, and to determine whether the patient meets the second blood sugar risk standard.

[0068] Beneficial effects of the present invention:

[0069] The present invention proposes a multi-parameter blood sugar risk warning method and system, which can more comprehensively evaluate the patient's blood sugar risk by comprehensively using a variety of blood sugar related data and two different risk prediction models, thereby improving the accuracy of the warning. The above technical solution realizes a multi-level evaluation of the patient's blood sugar risk through two risk predictions (the first and the second). This design helps to capture risks at different levels and provide more detailed health management advice to doctors and patients. Through timely data analysis and risk prediction, the method helps to detect patients' blood sugar abnormalities early, providing the possibility for early intervention and treatment, thereby reducing the risk of diabetes and its complications. Since the method can comprehensively consider the patient's various medical data information and individual differences, it helps to achieve more personalized blood sugar management and treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 is a flow chart of the method of the present invention;

[0071] Figure 2The system block diagram of the system of the present invention. DETAILED DESCRIPTION

[0072] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0073] The embodiment of the present invention proposes a blood sugar risk warning method based on multiple parameters, such as Figure 1 As shown, the multi-parameter-based blood sugar risk early warning method includes:

[0074] S1. Retrieving medical data information of a patient from a medical terminal, and obtaining first blood glucose data information and second blood glucose data information from the medical data information;

[0075] S2. Analyze the first blood glucose data information and the second blood glucose data information respectively using the first risk prediction model and the second risk prediction model to obtain a first risk prediction value and a second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information;

[0076] S3, performing a first blood glucose risk prediction for the patient according to the first risk prediction value, and determining whether the patient meets the first blood glucose risk standard;

[0077] S4. When the patient does not meet the first blood sugar risk standard, a second blood sugar risk prediction is performed on the patient using the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard.

[0078] The working principle of the above technical solution is as follows: First, retrieve the patient's comprehensive medical data information from the medical terminal. This information may include but is not limited to the patient's historical blood sugar records, physiological parameters, living habits, etc. Then, two types of blood sugar data information are specially extracted from these rich medical data information, namely the first blood sugar data information and the second blood sugar data information. These two types of data may represent blood sugar conditions under different test times, test conditions or related physiological states.

[0079] Next, two different risk prediction models (the first risk prediction model and the second risk prediction model) are used to conduct in-depth analysis on the first blood glucose data information and the second blood glucose data information. These two models may be built based on machine learning, statistical analysis or other advanced algorithms, and can mine risk information in blood glucose data from different angles. Through model analysis, two sets of risk prediction values ​​are obtained: the first risk prediction value and the second risk prediction value.

[0080] Based on the first risk prediction value, a preliminary blood sugar risk prediction is performed on the patient. This step is mainly to determine whether the patient has reached the blood sugar risk standard set for the first time. If so, it may mean that the patient needs immediate medical attention or intervention. If the patient does not meet the first blood sugar risk standard, the method will enter the second stage of prediction. In this stage, the method will comprehensively consider the first risk prediction value and the second risk prediction value to perform a more detailed risk assessment. The purpose of this step is to capture potential risks that may be overlooked in the initial prediction, thereby further improving the sensitivity and accuracy of the early warning system.

[0081] The effect of the above technical solution is: by comprehensively using a variety of blood sugar related data and two different risk prediction models, this method can more comprehensively evaluate the patient's blood sugar risk, thereby improving the accuracy of the early warning. The above technical solution realizes a multi-level assessment of the patient's blood sugar risk through two risk predictions (the first and the second). This design helps to capture different levels of risks and provide more detailed health management advice to doctors and patients. Through timely data analysis and risk prediction, this method helps to detect patients' blood sugar abnormalities early, providing the possibility for early intervention and treatment, thereby reducing the risk of diabetes and its complications. Because this method can comprehensively consider the patient's various medical data information and individual differences, it helps to achieve more personalized blood sugar management and treatment plans.

[0082] In one embodiment of the present invention, retrieving medical data information of a patient from the medical terminal, and acquiring first blood sugar data information and second blood sugar data information from the medical data information, includes:

[0083] S101, retrieving the patient's medical data information from the medical terminal;

[0084] S102, extracting multi-parameter blood glucose data information from the medical data information, wherein the multi-parameter blood glucose data information includes fasting blood glucose value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value, and data collection time sequence corresponding to the fasting blood glucose value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value;

[0085] S103, dividing the fasting blood glucose value in the multi-parameter blood glucose data information and its corresponding data collection time sequence into first blood glucose data information;

[0086] S104. Divide the glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value in the multi-parameter blood glucose data information, and the data collection time sequence corresponding to the glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value into second blood glucose data information.

[0087] The working principle of the above technical solution is as follows: First, the patient's medical data information is fully retrieved from the medical terminal. This information is a comprehensive record of the patient's health status, including various physiological parameters, test results, diagnosis records, etc. From the retrieved medical data information, special attention is paid to multi-parameter data related to blood sugar. These parameters include but are not limited to fasting blood sugar value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value. At the same time, the data collection time series corresponding to these data is extracted, which is an important time clue to understand data changes and patient health status.

[0088] The fasting blood glucose value and its corresponding data collection time series in the multi-parameter blood glucose data information are separated separately to form the first blood glucose data information. The fasting blood glucose value is an important indicator for evaluating blood glucose levels, and its time series changes are of great significance for understanding the patient's blood glucose fluctuations and possible risks. In addition to the fasting blood glucose value, other parameters (glycosylated hemoglobin, glycosylated albumin, insulin values, blood pressure values, and blood lipid values) and their corresponding data collection time series are divided into the second blood glucose data information. These parameters provide more dimensional information about the patient's blood glucose status, which helps to more comprehensively evaluate the patient's health status and risks.

[0089] The effect of the above technical solution is: through detailed classification and extraction of multi-parameter blood sugar data information, this technical solution realizes the refined processing of data, and provides a high-quality data foundation for subsequent data analysis and risk prediction. By comprehensively considering a variety of physiological parameters related to blood sugar, the solution can more comprehensively assess the patient's blood sugar risk, not only relying on a single fasting blood sugar value. By extracting the data collection time series corresponding to each parameter, the solution can analyze the dynamic changes of the patient's blood sugar status, providing strong support for the timely discovery of potential risks. Based on the data of refined processing and multi-dimensional evaluation, the subsequent risk prediction model is expected to improve the accuracy and reliability of the prediction, thereby better assisting medical decision-making.

[0090] In general, this technical solution lays a solid foundation for the accurate prediction of blood sugar risks and personalized management of patients through refined data processing and multi-dimensional risk assessment.

[0091] In one embodiment of the present invention, a first risk prediction model and a second risk prediction model are used to respectively analyze the first blood glucose data information and the second blood glucose data information to obtain a first risk prediction value and a second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information, including:

[0092] S201, inputting the first blood glucose data information into a first risk prediction model for risk prediction to obtain a first risk prediction value; wherein the first risk prediction model adopts a convolutional neural network model, and the model structure of the first risk prediction model includes a data input end, a TCN feature extraction network, an XGBoost time series prediction layer and a data output end;

[0093] S202. Input the second blood glucose data information into a second risk prediction model for risk prediction to obtain a second risk prediction value; the second risk prediction model adopts a binary logistic regression model.

[0094] The working principle of the above technical solution is: the working principle of the technical solution is to use two different risk prediction models (a first risk prediction model and a second risk prediction model) to independently analyze two sets of different blood sugar data information (first blood sugar data information and second blood sugar data information) to obtain their respective corresponding risk prediction values.

[0095] First, the first blood glucose data information (mainly fasting blood glucose value and its time series data) is input into the first risk prediction model. This model is a convolutional neural network model that combines deep learning technology and has complex data processing capabilities. The structure of the model includes a data input terminal, a TCN (Temporal Convolutional Networks) feature extraction network, an XGBoost time series prediction layer, and a data output terminal. The data enters the model through the input terminal, and the TCN network is responsible for extracting useful features from the time series data. These features are then passed to the XGBoost time series prediction layer for risk prediction analysis, and finally the first risk prediction value is output through the data output terminal.

[0096] Secondly, the second blood sugar data information (including glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value and their time series data) is input into the second risk prediction model. This model uses a binary logistic regression model, which is suitable for the case where the dependent variable is a binary classification and can predict the probability of blood sugar risk in patients based on the input of multiple physiological parameters. The model will perform statistical analysis on the input data and finally output the second risk prediction value.

[0097] The effect of the above technical solution is: this solution combines deep learning models (convolutional neural networks) and traditional statistical models (binary logistic regression), which can simultaneously process complex nonlinear features and linear relationships, and improve the accuracy and generalization of risk prediction. By processing two different sets of blood sugar data information with two different models, the solution can make more comprehensive use of the patient's medical data, thereby more accurately assessing blood sugar risks. The TCN network and XGBoost time series prediction layer in the first risk prediction model can effectively process time series data, capture the dynamic changes of blood sugar data, and provide richer information for risk prediction. Using two different types of models for analysis, the parameters and structure of the model can be adjusted according to the actual situation to adapt to the characteristics and needs of different patients and improve the personalization of the prediction.

[0098] In summary, this technical solution combines deep learning and traditional statistical methods to make full use of multi-dimensional and time series data, thereby improving the accuracy and flexibility of blood sugar risk prediction.

[0099] In one embodiment of the present invention, the first blood glucose data information is input into a first risk prediction model for risk prediction to obtain a first risk prediction value, including:

[0100] S2011, inputting the first blood glucose data information into a first risk prediction model;

[0101] S2012. Obtaining a blood glucose curve graph of the patient's corresponding future days through the first risk prediction model; wherein the future days have a value range of 15-30 days;

[0102] S2013, obtaining a first risk prediction value according to the blood glucose curve corresponding to the patient in the future days combined with the blood glucose standard range value; wherein the first risk prediction value is obtained by the following formula:

[0103]

[0104] Where R represents the first risk prediction value; X i represents the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model; n represents the number of input feature variables corresponding to the TCN feature extraction network; X up and X down They respectively represent the upper and lower limits of the healthy standard range of fasting blood sugar values.

[0105] The working principle of the above technical solution is: the working principle of the technical solution is to input the first blood sugar data information into the first risk prediction model, use the prediction ability of the model to generate the patient's blood sugar curve for a period of time in the future (15-30 days), and calculate the first risk prediction value in combination with the blood sugar standard range. The specific process is as follows:

[0106] The first blood glucose data information (mainly fasting blood glucose value and its time series data) is input into the first risk prediction model. This is the starting step of the prediction process and provides the necessary data basis for the model.

[0107] Using the processing power of the first risk prediction model (combining the TCN feature extraction network and the XGBoost time series prediction layer), the patient's fasting blood glucose values ​​for the next 15-30 days are predicted based on the input first blood glucose data information, and a blood glucose curve is generated accordingly. This curve shows the patient's blood glucose change trend over a period of time in the future.

[0108] Combining the generated blood glucose curve and the healthy standard range of fasting blood glucose (defined by Xup and Xdown), the first risk prediction value is calculated through a specific formula. This formula takes into account the relationship between each predicted fasting blood glucose value (Xi) output by the model and the standard range, as well as the number of input feature variables (n), thereby quantitatively evaluating the size of the patient's future blood glucose risk.

[0109] The effect of the above technical solution is: through the blood glucose curve chart for the next 15-30 days generated by the model, doctors and patients can intuitively understand the future blood glucose change trend, which is helpful to formulate intervention measures in advance. The calculation of the first risk prediction value provides a quantitative indicator for assessing the patient's future blood glucose risk, making risk management more accurate and objective. Since the model is based on the patient's individual data for prediction, it can provide personalized blood glucose risk prediction to meet the special needs of different patients. By combining the blood glucose standard range, this technical solution can assist doctors in formulating more reasonable treatment plans and timely adjust patients' medication and lifestyle habits to reduce blood glucose risks.

[0110] At the same time, by inputting blood sugar data into the first risk prediction model and combining it with the blood sugar curve chart of the next few days for prediction, the blood sugar fluctuations in the next 15-30 days can be estimated more accurately. This helps to identify potential risks of abnormal blood sugar in advance, so as to take early intervention measures. The first risk prediction model in this program not only takes into account the individual's blood sugar data, but also combines specific time ranges and individual characteristic variables to provide more personalized risk prediction results. This personalized assessment method can better meet the actual needs of patients and improve the accuracy and reliability of predictions. The risk assessment performed by the above formula compares the predicted results with the health standard range, making the assessment results more scientific and objective. This standardized assessment method helps to better quantify risks and facilitates clinicians or patients to understand and apply.

[0111] Overall, this technical solution provides doctors and patients with a powerful tool to better manage and control blood sugar levels and prevent complications by predicting future blood sugar trends and quantifying risk assessment.

[0112] In one embodiment of the present invention, the second blood glucose data information is input into a second risk prediction model for risk prediction to obtain a second risk prediction value, including:

[0113] S2021. Retrieve the binary logistic regression model and the multivariate logistic regression analysis table corresponding to glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value;

[0114] S2022. Extracting regression coefficients corresponding to glycated hemoglobin, glycated albumin, insulin value, blood pressure value, and blood lipid value from the multi-factor logistic regression analysis table;

[0115] S2023. Utilize the binary logistic regression model in combination with the regression coefficients corresponding to the glycated hemoglobin, glycated albumin, insulin value, blood pressure value, and blood lipid value to obtain a second risk prediction value, wherein the second risk prediction value is obtained by the following formula:

[0116]

[0117] Among them, P c represents the second risk prediction value; β0 represents the intercept term; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j Indicates the data value of the j-th second blood sugar data information.

[0118] The working principle of the above technical solution is as follows: first, the binary logistic regression model and the multi-factor logistic regression analysis table corresponding to glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value are retrieved. This step ensures that all the required data and models are ready for subsequent analysis.

[0119] The regression coefficients corresponding to each second blood glucose data information (glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value) were extracted from the multivariate logistic regression analysis table. These regression coefficients are key parameters in the binary logistic regression model, which reflect the influence of each variable on the dependent variable (i.e., risk).

[0120] Using the binary logistic regression model and the extracted regression coefficient, combined with the specific data value of the second blood glucose data information, the second risk prediction value Pc is calculated through a specific formula. This formula is a typical expression of a binary logistic regression model, where β0 is the intercept term, m represents the number of second blood glucose data information, and βj and Yj represent the regression coefficient and data value of the jth second blood glucose data information, respectively. In this way, the risk level of the patient based on these physiological parameters can be quantitatively evaluated.

[0121] The effect of the above technical solution is: by using the binary logistic regression model and specific physiological parameter data, the technical solution can provide an accurate and quantitative second risk prediction value, which helps doctors and patients to more accurately understand the risk level under the current health status. The solution comprehensively considers a variety of physiological parameters related to blood sugar risk (such as glycated hemoglobin, glycated albumin, etc.), so as to more comprehensively evaluate the patient's overall risk status rather than relying on a single indicator. Since the physiological parameter data of each patient is unique, the technical solution can provide each patient with personalized risk prediction and health management suggestions, thereby improving the pertinence and effectiveness of medical services.

[0122] On the other hand, by calling the binary logistic regression model and combining it with multi-factor analysis, it is possible to comprehensively analyze multi-dimensional blood sugar data such as glycosylated hemoglobin, glycosylated albumin, insulin values, blood pressure values, and blood lipid values. This multi-dimensional comprehensive analysis method helps to comprehensively evaluate the health status of patients and provide more accurate risk prediction results.

[0123] Through multi-factor Logistic regression analysis, multiple factors that affect blood sugar and related health indicators can be effectively extracted and analyzed, and their contribution to risk prediction can be calculated. This analysis method not only improves the efficiency of the prediction model, but also more accurately quantifies the impact of each factor on risk. The second risk prediction value calculated using the binary Logistic regression model can more accurately predict the risk that an individual may face in the future. This risk assessment method based on Logistic regression can significantly improve the accuracy of predicting patient risks in clinical applications due to its sensitivity and accuracy to data. The solution calculates the risk probability through a specific mathematical formula to convert complex multidimensional data into a risk probability value that is easy to understand and apply. This ability to quantify risk makes the prediction results more practical, facilitates the formulation of medical decisions, and facilitates the understanding and acceptance of patients.

[0124] In general, this technical solution achieves accurate prediction and quantitative assessment of patients' blood sugar risks by combining the binary logistic regression model and the patient's specific physiological parameter data, providing strong support for personalized medical services.

[0125] In one embodiment of the present invention, performing a first blood glucose risk prediction on a patient according to the first risk prediction value, and determining whether the patient meets the first blood glucose risk standard, includes:

[0126] S301, comparing the first risk prediction value with a preset first risk threshold;

[0127] S302: When the first risk prediction value exceeds a preset first risk threshold, it is determined that the patient has reached the first blood sugar risk standard, and a blood sugar risk warning is issued;

[0128] S303: When the first risk prediction value does not exceed the preset first risk threshold, it is determined that the patient does not meet the first blood sugar risk standard.

[0129] The working principle of the above technical solution is: first, a first risk threshold is preset as a standard for judging the patient's blood sugar risk.

[0130] The first risk prediction value is compared with the preset first risk threshold value to evaluate whether the patient's blood sugar risk has reached a level that requires attention through numerical analysis.

[0131] If the first risk prediction value exceeds the preset first risk threshold, the system will determine that the patient has reached the first blood sugar risk standard and trigger a blood sugar risk warning. This is to detect potential high blood sugar risks in a timely manner so that early intervention can be carried out.

[0132] On the contrary, if the first risk prediction value does not exceed the preset first risk threshold, the system will determine that the patient has not reached the first blood sugar risk standard, which means that the current blood sugar risk of the patient is within an acceptable range and there is no need to trigger an early warning.

[0133] The effect of the above technical solution is: by setting a risk threshold and comparing it with the real-time predicted risk value, the technical solution can detect abnormalities in the patient's blood sugar in a timely manner, thereby allowing medical staff and patients to take necessary measures to prevent or control the risks brought by high blood sugar. Risk judgment based on individualized risk prediction values ​​improves the accuracy and personalization of blood sugar management. Unlike the traditional fixed threshold judgment method, this comparison based on individual predicted values ​​and risk thresholds can better reflect the patient's actual risk status. Through reasonable threshold setting and comparison with predicted values, the technical solution aims to reduce false alarms (i.e., misidentifying low-risk patients as high-risk) and missed alarms (i.e., failing to identify truly high-risk patients in a timely manner), thereby improving the reliability and effectiveness of the blood sugar risk warning system.

[0134] In general, this technical solution achieves timely warning and precise management of patients' blood sugar risks by comparing individualized risk prediction values ​​with preset risk thresholds, which helps to improve patients' health levels and the quality of medical services.

[0135] In one embodiment of the present invention, when the patient does not meet the first blood sugar risk standard, a second blood sugar risk prediction is performed on the patient using the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard, including:

[0136] S401, when the patient does not meet the first blood sugar risk standard, retrieving a second risk prediction value;

[0137] S402, comparing the second risk prediction value with a preset second risk threshold;

[0138] S403: when the second risk prediction value exceeds a preset second risk threshold, using the first comprehensive risk model in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard;

[0139] S404: When the second risk prediction value does not exceed the preset second risk threshold, a second comprehensive risk model is used to combine the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard.

[0140] The working principle of the above technical solution is as follows: First, confirm that the patient does not meet the first blood sugar risk standard, which is the premise for the second blood sugar risk prediction. The second risk prediction value is retrieved and compared with the preset second risk threshold. This step is to make a preliminary assessment of the patient's blood sugar risk from another dimension (based on different physiological parameters and data models).

[0141] If the second risk prediction value exceeds the second risk threshold, it means that from the perspective of the second risk prediction model, the patient has a higher blood sugar risk. At this time, the first comprehensive risk model is used to combine the first risk prediction value and the second risk prediction value to conduct a more comprehensive risk assessment to determine whether the patient actually meets the second blood sugar risk standard. If the second risk prediction value does not exceed the second risk threshold, the second comprehensive risk model is used to conduct a risk assessment. This step takes into account that even if a single risk prediction model does not show a high risk, the results of multiple models combined may reveal potential risks.

[0142] The effect of the above technical solution is: by combining the first risk prediction value and the second risk prediction value, the technical solution provides a method for evaluating the patient's blood sugar risk from multiple angles, which increases the comprehensiveness and accuracy of risk assessment. According to the comparison result between the second risk prediction value and the second risk threshold, different comprehensive risk models (the first comprehensive risk model or the second comprehensive risk model) can be flexibly selected and applied, which reflects the refinement and personalization of risk assessment. Through two risk assessments, the technical solution can more sensitively capture subtle changes in the patient's blood sugar risk, thereby issuing early warnings, which is conducive to early intervention and prevention of adverse events.

[0143] In general, this technical solution improves the accuracy and sensitivity of blood sugar risk warning through multi-dimensional and multi-model comprehensive evaluation, which helps to provide patients with more accurate and personalized health management services.

[0144] In one embodiment of the present invention, when the second risk prediction value exceeds a preset second risk threshold, the first comprehensive risk model is used in combination with the first risk prediction value and the second risk prediction value to determine whether the patient reaches the second blood sugar risk standard, including:

[0145] S4031. When the second risk prediction value exceeds a preset second risk threshold, a weight value is obtained using the second risk prediction value and the second risk threshold, wherein the weight value is obtained by the following formula:

[0146]

[0147] Where w(R) represents the weight value; R represents the first risk prediction value; R y represents the first risk threshold; P represents the second risk prediction value; P y represents the second risk threshold;

[0148] S4032. Obtain a first comprehensive risk model using the weight value; wherein the structure of the first comprehensive risk model is as follows:

[0149]

[0150] Among them, P 01 represents the first comprehensive risk parameter obtained by the first comprehensive risk model; β0 represents the intercept term; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j A data value representing the j-th second blood sugar data information;

[0151] S4033, substituting the first risk prediction value and the second blood glucose data information into a first comprehensive risk model to obtain a first comprehensive risk parameter;

[0152] S4034, comparing the first comprehensive risk parameter with a preset first comprehensive parameter threshold;

[0153] S4035. When the first comprehensive risk parameter exceeds the preset first comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard, and a blood sugar risk warning is issued.

[0154] The working principle of the above technical solution is: when the second risk prediction value exceeds the preset second risk threshold, the second risk prediction value and the second risk threshold are first used, combined with the first risk prediction value and the first risk threshold, to calculate the weight value through a given formula. This weight value reflects the importance of the two risk assessment results and provides a basis for subsequent comprehensive risk assessment. Using the calculated weight value, combined with the second blood glucose data information and the corresponding regression coefficient, a first comprehensive risk model is constructed. This model comprehensively considers a variety of blood glucose data information to more comprehensively assess the patient's blood glucose risk.

[0155] Substitute the first risk prediction value and the second blood glucose data information into the first comprehensive risk model, and calculate the first risk comprehensive parameter. This parameter is a risk assessment indicator that integrates various information. Compare the calculated first risk comprehensive parameter with the preset first comprehensive parameter threshold to determine whether the patient's blood glucose risk has reached the second blood glucose risk standard. If the first risk comprehensive parameter exceeds the preset first comprehensive parameter threshold, it is determined that the patient has reached the second blood glucose risk standard, and a blood glucose risk warning is issued in a timely manner.

[0156] The effect of the above technical solution is: by combining the first risk prediction value and the second blood glucose data information, and considering the weight value, the technical solution can provide a more comprehensive and comprehensive blood glucose risk assessment method. The use of weight values ​​to adjust the influence of different risk factors in the comprehensive assessment increases the flexibility and accuracy of the assessment. At the same time, by comparing the first risk comprehensive parameter with the preset threshold, it is possible to more accurately determine whether the patient meets the blood glucose risk standard. Once it is determined that the patient meets the second blood glucose risk standard, the system will immediately issue a blood glucose risk warning, which will help medical staff and patients take timely measures to control blood glucose levels and reduce potential health risks.

[0157] On the other hand, by combining the results of the first risk prediction model and the second risk prediction model, a multiple risk assessment method is provided. This process can help to more accurately determine whether the patient has or is about to reach a dangerous blood sugar level.

[0158] The scheme introduces a weight coefficient w(R) to adjust and weigh different risk prediction results. This can dynamically adjust the focus of risk assessment to ensure that the system pays enough attention when the predicted value exceeds the threshold, thereby improving the accuracy and reliability of the assessment. By combining the results of different prediction models to form a comprehensive risk model, it can provide patients with more personalized risk predictions to meet the needs of different individuals, further improving the applicability of the scheme. Ultimately, these risk prediction results are used to determine whether the patient has reached the secondary blood sugar risk standard, and further risk management and intervention are carried out accordingly. This provides important support for clinical decision-making and can better protect the health of patients.

[0159] At the same time, by adjusting the weight w(R) and combining the results of different risk models, the mathematical model can effectively balance the output of different models and improve the accuracy of overall risk prediction. The weight coefficient w(R) and the comprehensive risk formula in the mathematical model can adapt to the risk assessment needs of different patients and different situations, making the model highly flexible. By combining the information of the first and second risk prediction models, the mathematical model can integrate multi-source data and improve the ability to comprehensively assess the patient's future blood sugar risk. This comprehensive ability ensures that the model can still make reliable risk assessments when faced with complex health data. The model can be dynamically adjusted according to real-time risk assessment results. When the second risk prediction value exceeds the set threshold, the model can automatically adjust the weight to enhance the sensitivity and accuracy of the prediction results.

[0160] In summary, this technical solution improves the accuracy and reliability of risk prediction through the combination of multiple models and weight adjustment mechanism. The mathematical model improves its adaptability to complex health data through flexible weight adjustment and data integration in performance indicators, providing support for more accurate clinical decision-making. At the same time, this technical solution improves the accuracy and timeliness of blood sugar risk warning by constructing a comprehensive risk assessment model and combining weight values ​​and multiple blood sugar data information to conduct a comprehensive risk assessment.

[0161] In one embodiment of the present invention, when the second risk prediction value does not exceed the preset second risk threshold, a second comprehensive risk model is used to combine the first risk prediction value and the second risk prediction value to determine whether the patient reaches the second blood sugar risk standard, including:

[0162] S4041. When the second risk prediction value does not exceed the preset second risk threshold, retrieve the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network;

[0163] S4042: Obtain a complex evaluation function using the first risk prediction value and the input feature variables corresponding to the TCN feature extraction network, wherein the structure of the complex evaluation function is as follows:

[0164]

[0165] Among them, f(R, X) represents the complex evaluation function; n represents the number of input feature variables corresponding to the TCN feature extraction network; X i represents the predicted fasting blood glucose value outputted by the i-th first risk prediction model; Y j represents the data value corresponding to the j-th second blood glucose data information; α i represents the weight coefficient between the predicted fasting blood glucose value outputted by the i-th first risk prediction model and the first risk prediction value R; θ ij represents the average value of the predicted fasting blood glucose value outputted by the i-th first risk prediction model and the j-th second blood glucose data information and the weight value of the first risk prediction value R; R represents the first risk prediction value; m represents the number of second blood glucose data information;

[0166] S4043: Obtain an influence correlation function using the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network, wherein the influence correlation function has the following structure:

[0167]

[0168] Among them, g(R, X) represents the influence correlation function; n represents the number of input feature variables corresponding to the TCN feature extraction network; X i represents the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model; R represents the first risk prediction value; δ represents the interaction coefficient between the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model and the first risk prediction value R, indicating the degree of influence of the characteristic variable and the first risk prediction value R on the second risk prediction value, and the interaction term reflects how the performance of a specific feature at different risk levels affects the risk prediction;

[0169] S4044. Obtain a second comprehensive risk model using the complex evaluation function and the impact association function, wherein the structure of the second comprehensive risk model is as follows:

[0170]

[0171] Among them, P 02 represents the second comprehensive risk parameter obtained by the second comprehensive risk model; β 01 represents the intercept term, which means the value of logit(p) of the binary logistic regression model when the predicted fasting blood glucose value and the first risk prediction value of the output corresponding to the i-th first risk prediction model of all characteristic variables are both 0; m represents the number of second blood glucose data information; β jY represents the regression coefficient of the j-th second blood glucose data information; j represents the data value of the j-th second blood sugar data information; β R represents the regression coefficient of the first risk prediction value; γ represents the quadratic coefficient of the first risk prediction value, which is used to capture the nonlinear effect of the first risk prediction value R on logit(p). It represents the influence of the square of the first risk prediction value R on the final predicted risk, reflecting the impact of the nonlinear effect on the model when the risk value is high or low; g(R, X) represents the impact correlation function; f(R, X) represents the complex evaluation function; λ represents the adjustment coefficient corresponding to the complex evaluation function, and the range of the adjustment coefficient is [-1,1];

[0172] S4045, substituting the first risk prediction value and the second blood glucose data information into a second comprehensive risk model to obtain a second comprehensive risk parameter;

[0173] S4046. Compare the second comprehensive risk parameter with a preset second comprehensive parameter threshold;

[0174] S4047. When the second comprehensive risk parameter exceeds the preset second comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard, and a blood sugar risk warning is issued.

[0175] The working principle of the above technical solution is: when the second risk prediction value does not exceed the preset threshold, the system will retrieve the first risk prediction value and the input feature variables corresponding to the TCN feature extraction network. These data provide the basis for subsequent comprehensive risk assessment.

[0176] A complex evaluation function is constructed using the first risk prediction value and the input feature variables of the TCN feature extraction network. This function comprehensively considers multiple risk factors, including the predicted fasting blood glucose value, the second blood glucose data information, etc., as well as the weight relationship between them and the first risk prediction value.

[0177] In order to capture the interaction between the characteristic variables and the first risk prediction value, an impact association function is constructed. This function reflects how specific characteristics affect risk prediction at different risk levels. Combining the complex evaluation function and the impact association function, a second comprehensive risk model is constructed. This model not only considers the linear relationship, but also captures the nonlinear impact by introducing the quadratic term coefficient of the first risk prediction value.

[0178] Substitute the first risk prediction value and the second blood glucose data information into the second comprehensive risk model to calculate the second risk comprehensive parameter. Compare this parameter with the preset second comprehensive parameter threshold. If the second risk comprehensive parameter exceeds the preset threshold, it is determined that the patient has reached the second blood glucose risk standard and triggers a blood glucose risk warning.

[0179] The effect of the above technical solution is: by integrating multiple risk factors and characteristic variables, this technical solution provides a comprehensive blood sugar risk assessment method that is more accurate than a single prediction model. By affecting the correlation function, the solution can capture the interaction between the characteristic variables and the risk prediction value, thereby improving the accuracy of risk assessment. By introducing the quadratic term coefficient, the model can capture the nonlinear impact of the risk value, making the assessment results closer to the actual situation. Once the second risk comprehensive parameter exceeds the preset threshold, the system will immediately issue an early warning, which will help to take timely intervention measures to reduce blood sugar risks.

[0180] At the same time, by introducing complex evaluation functions and influence correlation functions, and combining the TCN feature extraction network to process the input feature variables, the scheme can make a more refined and multi-level assessment of blood sugar risk. This method can capture more subtle changes and influencing factors in risk assessment, thereby improving the accuracy of prediction. By comparing the second risk comprehensive parameter with the preset threshold, the scheme can dynamically adjust the risk assessment and early warning mechanism. This dynamic adjustment mechanism ensures that in practical applications, risks can be judged and responded to in a timely and flexible manner based on real-time data, reducing the possibility of misjudgment and missed judgment. In the judgment and early warning process of the second blood sugar risk standard, the scheme considers multiple complex feature variables and their interactive effects, so it can provide individuals with more personalized risk early warning services. This personalized early warning method can better help patients take preventive measures early to avoid blood sugar out of control. By substituting the first risk prediction value and the second blood sugar data information into the second comprehensive risk model, the scheme effectively integrates multi-dimensional data and provides comprehensive risk assessment and prediction. This data integration capability enhances the overall prediction performance of the model and provides stronger support for patients' health management.

[0181] On the other hand, through the complex evaluation function f (R, X) and the influence correlation function g (R, X), the mathematical model can capture and analyze more dimensional features and influencing factors, thereby improving the overall prediction ability of the model. The introduction of this complex function makes the model more sensitive to potential risks. The second comprehensive risk model in the mathematical model combines multiple complex functions with basic regression analysis, which can not only efficiently calculate the second risk comprehensive parameters, but also accurately determine whether the patient meets the secondary blood sugar risk standard. This efficiency ensures the practicality of the model in application and can provide timely risk warnings.

[0182] Through the interaction of features calculated by the complex evaluation function f (R,X) and the influence correlation function g (R,X), the mathematical model can deeply analyze the mutual influence between different feature variables. This sophisticated feature analysis capability enables the model to maintain high accuracy and robustness when facing complex data. The mathematical model has the ability to respond and adjust dynamically. It can adjust the weights and parameters in time according to the actual risk assessment results to optimize the model's predictive performance. This dynamic adjustment capability improves the adaptability of the model, allowing it to work effectively in different situations.

[0183] In summary, this technical solution improves the precision and personalization of risk assessment by introducing complex evaluation functions and impact correlation functions combined with TCN feature extraction network. The mathematical model performs well in performance indicators by enhancing prediction capabilities, feature interaction analysis and dynamic adjustment mechanisms, providing strong support for accurate prediction and early warning of blood sugar risks. At the same time, this technical solution improves the accuracy and timeliness of blood sugar risk assessment by building a complex comprehensive risk assessment model that integrates multiple risk factors and characteristic variables.

[0184] The embodiment of the present invention proposes a blood sugar risk early warning system based on multiple parameters, such as Figure 2 As shown, the multi-parameter-based blood sugar risk early warning system includes:

[0185] a blood sugar data information acquisition module, used to retrieve the patient's medical data information from the medical terminal, and acquire first blood sugar data information and second blood sugar data information from the medical data information;

[0186] a risk prediction value acquisition module, used to perform data analysis on the first blood glucose data information and the second blood glucose data information respectively using the first risk prediction model and the second risk prediction model, and acquire the first risk prediction value and the second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information;

[0187] A first blood sugar risk standard determination module, used to perform a first blood sugar risk prediction for the patient according to the first risk prediction value, and determine whether the patient meets the first blood sugar risk standard;

[0188] The second blood sugar risk standard determination module is used to perform a second blood sugar risk prediction on the patient using the first risk prediction value and the second risk prediction value when the patient does not meet the first blood sugar risk standard, and to determine whether the patient meets the second blood sugar risk standard.

[0189] The working principle of the above technical solution is as follows: First, retrieve the patient's comprehensive medical data information from the medical terminal. This information may include but is not limited to the patient's historical blood sugar records, physiological parameters, living habits, etc. Then, two types of blood sugar data information are specially extracted from these rich medical data information, namely the first blood sugar data information and the second blood sugar data information. These two types of data may represent blood sugar conditions under different test times, test conditions or related physiological states.

[0190] Next, two different risk prediction models (the first risk prediction model and the second risk prediction model) are used to conduct in-depth analysis on the first blood glucose data information and the second blood glucose data information. These two models may be built based on machine learning, statistical analysis or other advanced algorithms, and can mine risk information in blood glucose data from different angles. Through model analysis, two sets of risk prediction values ​​are obtained: the first risk prediction value and the second risk prediction value.

[0191] Based on the first risk prediction value, a preliminary blood sugar risk prediction is performed on the patient. This step is mainly to determine whether the patient has reached the blood sugar risk standard set for the first time. If so, it may mean that the patient needs immediate medical attention or intervention. If the patient does not meet the first blood sugar risk standard, the method will enter the second stage of prediction. In this stage, the method will comprehensively consider the first risk prediction value and the second risk prediction value to perform a more detailed risk assessment. The purpose of this step is to capture potential risks that may be overlooked in the initial prediction, thereby further improving the sensitivity and accuracy of the early warning system.

[0192] The effect of the above technical solution is: by comprehensively using a variety of blood sugar related data and two different risk prediction models, this method can more comprehensively evaluate the patient's blood sugar risk, thereby improving the accuracy of the early warning. The above technical solution realizes a multi-level assessment of the patient's blood sugar risk through two risk predictions (the first and the second). This design helps to capture different levels of risks and provide more detailed health management advice to doctors and patients. Through timely data analysis and risk prediction, this method helps to detect patients' blood sugar abnormalities early, providing the possibility for early intervention and treatment, thereby reducing the risk of diabetes and its complications. Because this method can comprehensively consider the patient's various medical data information and individual differences, it helps to achieve more personalized blood sugar management and treatment plans.

[0193] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A multi-parameter blood sugar risk warning method, characterized in that: The multi-parameter-based blood sugar risk early warning method includes: Retrieving medical data information of the patient from the medical terminal, and acquiring first blood sugar data information and second blood sugar data information from the medical data information; Using the first risk prediction model and the second risk prediction model to respectively analyze the first blood sugar data information and the second blood sugar data information, and obtain a first risk prediction value and a second risk prediction value corresponding to the first blood sugar data information and the second blood sugar data information; Performing a first blood glucose risk prediction on the patient according to the first risk prediction value, and determining whether the patient meets the first blood glucose risk standard; When the patient does not meet the first blood sugar risk standard, a second blood sugar risk prediction is performed on the patient using the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard. Wherein, when the patient does not meet the first blood sugar risk standard, a second risk prediction value is retrieved; the second risk prediction value is compared with a preset second risk threshold, and the following steps are performed; S1: when the second risk prediction value exceeds a preset second risk threshold, the first comprehensive risk model is used in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard; S2: when the second risk prediction value does not exceed the preset second risk threshold, using the second comprehensive risk model in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard; Wherein, operation step S1 includes: When the second risk prediction value exceeds a preset second risk threshold, a weight value is obtained using the first risk prediction value and the second risk prediction value, and a first comprehensive risk model is obtained using the weight value; Substituting the first risk prediction value and the second blood glucose data information into a first comprehensive risk model to obtain a first comprehensive risk parameter; comparing the first comprehensive risk parameter with a preset first comprehensive parameter threshold; When the first risk comprehensive parameter exceeds the preset first comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard, and a blood sugar risk warning is issued; Wherein, operation step S2 includes: When the second risk prediction value does not exceed the preset second risk threshold, calling the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network; Obtaining a complex evaluation function using the first risk prediction value and the input feature variables corresponding to the TCN feature extraction network, The impact correlation function is obtained by using the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network, Obtaining a second comprehensive risk model using the complex evaluation function and the impact association function; Substituting the first risk prediction value and the second blood glucose data information into a second comprehensive risk model to obtain a second comprehensive risk parameter; The second risk comprehensive parameter is compared with a preset second comprehensive parameter threshold; when the second risk comprehensive parameter exceeds the preset second comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard, and a blood sugar risk warning is issued.

2. The multi-parameter blood sugar risk early warning method according to claim 1, characterized in that: Retrieving medical data information of the patient from the medical terminal, and acquiring first blood sugar data information and second blood sugar data information from the medical data information, including: Retrieving the patient's medical data information from the medical terminal; Extracting multi-parameter blood sugar data information from the medical data information, wherein the multi-parameter blood sugar data information includes fasting blood sugar value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value, and data collection time sequence corresponding to the fasting blood sugar value, glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value; dividing the fasting blood glucose value in the multi-parameter blood glucose data information and its corresponding data collection time sequence into first blood glucose data information; The glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value in the multi-parameter blood glucose data information, as well as the data collection timing corresponding to the glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value are divided into the second blood glucose data information.

3. The multi-parameter blood sugar risk early warning method according to claim 1, characterized in that: Using the first risk prediction model and the second risk prediction model to respectively analyze the first blood glucose data information and the second blood glucose data information to obtain a first risk prediction value and a second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information, including: Inputting the first blood glucose data information into a first risk prediction model for risk prediction to obtain a first risk prediction value; wherein the first risk prediction model adopts a convolutional neural network model, and the model structure of the first risk prediction model includes a data input end, a TCN feature extraction network, an XGBoost time series prediction layer and a data output end; The second blood glucose data information is input into a second risk prediction model for risk prediction to obtain a second risk prediction value; the second risk prediction model adopts a binary logistic regression model.

4. The multi-parameter blood sugar risk early warning method according to claim 3, characterized in that: Inputting the first blood glucose data information into a first risk prediction model to perform risk prediction to obtain a first risk prediction value includes: inputting the first blood glucose data information into a first risk prediction model; Obtaining a blood glucose curve graph of the patient's corresponding future days through the first risk prediction model; wherein the future days have a value range of 15-30 days; The first risk prediction value is obtained according to the blood glucose curve corresponding to the patient in the future days combined with the blood glucose standard range value; wherein the first risk prediction value is obtained by the following formula: Where R represents the first risk prediction value; X i represents the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model; n represents the number of input feature variables corresponding to the TCN feature extraction network; X up and X down They respectively represent the upper and lower limits of the healthy standard range of fasting blood sugar values.

5. The multi-parameter blood sugar risk early warning method according to claim 3, characterized in that: Inputting the second blood glucose data information into a second risk prediction model for risk prediction to obtain a second risk prediction value includes: Retrieve the binary logistic regression model and the multivariate logistic regression analysis table corresponding to glycosylated hemoglobin, glycosylated albumin, insulin value, blood pressure value and blood lipid value; Extract the regression coefficients corresponding to glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value from the multi-factor logistic regression analysis table; The binary logistic regression model is combined with the regression coefficients corresponding to glycated hemoglobin, glycated albumin, insulin value, blood pressure value and blood lipid value to obtain a second risk prediction value, wherein the second risk prediction value is obtained by the following formula: Among them, P c represents the second risk prediction value; β0 represents the intercept term; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j Indicates the data value of the j-th second blood sugar data information.

6. The multi-parameter blood sugar risk early warning method according to claim 1, characterized in that: Performing a first blood glucose risk prediction for the patient according to the first risk prediction value to determine whether the patient meets the first blood glucose risk standard includes: Comparing the first risk prediction value with a preset first risk threshold; When the first risk prediction value exceeds a preset first risk threshold, it is determined that the patient has reached the first blood sugar risk standard, and a blood sugar risk warning is issued; When the first risk prediction value does not exceed the preset first risk threshold, it is determined that the patient does not meet the first blood sugar risk standard.

7. The multi-parameter blood sugar risk early warning method according to claim 1, characterized in that: The weight value is obtained by the following formula: Wherein, w(R) represents the weight value; R represents the first risk prediction value; R y represents the first risk threshold; P represents the second risk prediction value; P y represents the second risk threshold; The first comprehensive risk model is obtained by using the weight value; wherein the structure of the first comprehensive risk model is as follows: Among them, P 01 represents the first comprehensive risk parameter obtained by the first comprehensive risk model; β0 represents the intercept term; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j Indicates the data value of the j-th second blood sugar data information.

8. The multi-parameter blood sugar risk early warning method according to claim 1, characterized in that: The complex evaluation function structure is as follows: Where f(R, X) represents the complex evaluation function; n represents the number of input feature variables corresponding to the TCN feature extraction network; X i represents the predicted fasting blood glucose value outputted by the i-th first risk prediction model; Y j represents the data value corresponding to the j-th second blood glucose data information; α i represents the weight coefficient between the predicted fasting blood glucose value outputted by the i-th first risk prediction model and the first risk prediction value R; θ ij represents the average value of the predicted fasting blood glucose value outputted by the i-th first risk prediction model and the j-th second blood glucose data information and the weight value of the first risk prediction value R; R represents the first risk prediction value; m represents the number of second blood glucose data information; The influence association function is obtained by using the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network, wherein the structure of the influence association function is as follows: Among them, g(R, X) represents the influence correlation function; n represents the number of input feature variables corresponding to the TCN feature extraction network; X i represents the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model; R represents the first risk prediction value; δ represents the interaction coefficient between the predicted fasting blood glucose value of the output corresponding to the i-th first risk prediction model and the first risk prediction value R; The complex evaluation function and the impact association function are used to obtain a second comprehensive risk model, wherein the structure of the second comprehensive risk model is as follows: Among them, P 02 represents the second comprehensive risk parameter obtained by the second comprehensive risk model; β 01 represents the intercept term, which means the value of logit(p) of the binary logistic regression model when the predicted fasting blood glucose value and the first risk prediction value of the output corresponding to the i-th first risk prediction model of all characteristic variables are 0; m represents the number of second blood glucose data information; β j Y represents the regression coefficient of the j-th second blood glucose data information; j represents the data value of the j-th second blood sugar data information; β R represents the regression coefficient of the first risk prediction value; γ represents the quadratic term coefficient of the first risk prediction value; g(R, X) represents the influence correlation function; f(R, X) represents the complex evaluation function; λ represents the adjustment coefficient corresponding to the complex evaluation function, and the value range of the adjustment coefficient is [-1,1].

9. A multi-parameter blood sugar risk warning system, characterized in that: The multi-parameter-based blood sugar risk early warning system includes: A blood sugar data information acquisition module, used to retrieve the patient's medical data information from the medical terminal, and acquire the first blood sugar data information and the second blood sugar data information from the medical data information; a risk prediction value acquisition module, used to perform data analysis on the first blood glucose data information and the second blood glucose data information respectively using the first risk prediction model and the second risk prediction model, and acquire the first risk prediction value and the second risk prediction value corresponding to the first blood glucose data information and the second blood glucose data information; A first blood sugar risk standard determination module, used to perform a first blood sugar risk prediction for the patient according to the first risk prediction value, and determine whether the patient meets the first blood sugar risk standard; A second blood sugar risk standard determination module is used to perform a second blood sugar risk prediction on the patient using the first risk prediction value and the second risk prediction value when the patient does not meet the first blood sugar risk standard, and determine whether the patient meets the second blood sugar risk standard; Wherein, when the patient does not meet the first blood sugar risk standard, a second risk prediction value is retrieved; the second risk prediction value is compared with a preset second risk threshold, and the following steps are performed; S1: when the second risk prediction value exceeds a preset second risk threshold, the first comprehensive risk model is used in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard; S2: when the second risk prediction value does not exceed the preset second risk threshold, using the second comprehensive risk model in combination with the first risk prediction value and the second risk prediction value to determine whether the patient meets the second blood sugar risk standard; Wherein, operation step S1 includes: When the second risk prediction value exceeds a preset second risk threshold, a weight value is obtained using the first risk prediction value and the second risk prediction value, and a first comprehensive risk model is obtained using the weight value; Substituting the first risk prediction value and the second blood glucose data information into a first comprehensive risk model to obtain a first comprehensive risk parameter; comparing the first comprehensive risk parameter with a preset first comprehensive parameter threshold; When the first risk comprehensive parameter exceeds the preset first comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard, and a blood sugar risk warning is issued; Wherein, operation step S2 includes: When the second risk prediction value does not exceed the preset second risk threshold, calling the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network; Obtaining a complex evaluation function using the first risk prediction value and the input feature variables corresponding to the TCN feature extraction network, The impact correlation function is obtained by using the first risk prediction value and the input feature variable corresponding to the TCN feature extraction network, Obtaining a second comprehensive risk model using the complex evaluation function and the impact association function; Substituting the first risk prediction value and the second blood glucose data information into a second comprehensive risk model to obtain a second comprehensive risk parameter; The second risk comprehensive parameter is compared with a preset second comprehensive parameter threshold; when the second risk comprehensive parameter exceeds the preset second comprehensive parameter threshold, it is determined that the patient has reached the second blood sugar risk standard, and a blood sugar risk warning is issued.

Images