Blood glucose data prediction processing method and system for diabetic patient

Abstract

The invention discloses a blood glucose data prediction processing method and system for a diabetic patient. The method comprises the following operation steps: acquiring historical data of the diabetic patient; performing data preprocessing on the historical data to obtain a preprocessed historical data set; constructing a multiple linear regression prediction model through the historical data set; and predicting the blood glucose level of the diabetic patient based on the multiple linear regression prediction model. A multivariate linear regression prediction model is constructed by taking a multivariate linear regression model as a framework through historical data, so that the prediction of the blood glucose level of a diabetic patient is improved, and the prediction error is reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a method and system for predicting and processing blood sugar data of diabetic patients. Background Art

[0002] Diabetes is a chronic disease that affects millions of people worldwide and is characterized by a long-term state of high blood sugar. It can lead to serious complications such as cardiovascular disease, kidney failure, blindness and nerve damage. Therefore, blood sugar control is crucial for patients with diabetes. Management of blood sugar usually requires continuous monitoring and treatment to reduce the risk of complications for patients.

[0003] Traditional blood glucose monitoring methods rely on self-measurement of blood glucose by patients, laboratory tests, and clinical examinations. However, these methods have certain limitations, such as the reliance on frequent measurements by patients, time interval restrictions, and the prediction models relied on by doctors may not reflect blood glucose changes in real time and dynamically.

[0004] In order to better predict blood sugar changes and assist diabetic patients with personalized management, the application of artificial intelligence (AI) technology in diabetes monitoring and treatment has received widespread attention in recent years. By analyzing historical blood sugar data and related physiological indicator data, AI can build a more accurate blood sugar prediction model and provide personalized treatment plans, thereby helping diabetic patients manage blood sugar more effectively. However, building a prediction model directly from raw historical data may face the challenge of reducing prediction bias caused by data quality issues. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for predicting and processing blood sugar data of diabetic patients, which solves the above-mentioned technical problems pointed out in the prior art.

[0006] The present invention provides a method for predicting and processing blood sugar data of a diabetic patient, comprising the following steps:

[0007] Obtain historical data of diabetic patients;

[0008] Performing data preprocessing on the historical data to obtain a preprocessed historical data set;

[0009] A multiple linear regression prediction model is constructed using the historical data set; and the blood glucose level of the diabetic patient is predicted based on the multiple linear regression prediction model.

[0010] Preferably, the historical data includes: historical blood sugar values ​​and physiological index data.

[0011] Preferably, a multivariate linear regression prediction model is constructed by using the historical data set; and the blood glucose level of the diabetic patient is predicted based on the multivariate linear regression prediction model. The specific operation steps are as follows:

[0012] Using a multiple linear regression model, the historical blood sugar values ​​and physiological index data in the historical data set are used as input independent variables for regression analysis to form a feature matrix X;

[0013] Each row of the feature matrix represents a data sample, and each column represents a feature;

[0014] A set of features in the feature matrix: x1, x2,…, x n ; Use the input target data to predict the blood sugar value y, the expression is:

[0015] y=β0+β1·X1+β2·X2+…+β n ·X n +∈;

[0016] Where y is the predicted blood glucose value;

[0017] β0 is the intercept term;

[0018] β0,β1,…,β n ; is the regression coefficient;

[0019] x1,x2,…,x n is the input feature;

[0020] ∈ is the error term;

[0021] The regression coefficients are estimated using the least squares method using the historical data set:

[0022]

[0023] Where X is the feature matrix;

[0024] X T is the transpose of the feature matrix X;

[0025] X T X) -1 are feature matrices X and X T The inverse matrix of multiplication;

[0026] y is the predicted blood glucose value;

[0027] is the estimated value of the regression coefficient;

[0028] The regression coefficients are obtained by fitting the multivariate linear regression model through the historical data set. Thus, a multiple linear regression prediction model is obtained;

[0029] The multiple linear regression prediction model is optimized and predicted by cross-validation to obtain the final multiple linear regression prediction model.

[0030] Preferably, the basic form of the multivariate linear regression prediction model is:

[0031]

[0032] In the formula, is the real blood sugar value;

[0033] y is the predicted blood glucose value;

[0034] X is the feature matrix, which contains all the independent variables;

[0035] β is the regression coefficient, which is the parameter that needs to be estimated in regression analysis and is usually obtained through data fitting;

[0036] ∑ is the error term, which represents the difference between the model prediction value and the true value.

[0037] Preferably, the blood glucose level of the diabetic patient is predicted based on the multivariate linear regression prediction model to obtain the latest real-time data of the diabetic patient; the blood glucose level of the diabetic patient is predicted based on the multivariate linear regression prediction model to obtain the blood glucose level of the diabetic patient.

[0038] Preferably, the multivariate linear regression prediction model is optimized and predicted by cross validation to obtain the final multivariate linear regression prediction model. The specific operation steps are as follows:

[0039] The historical data set is evenly divided into k subsets by using k-fold cross validation, and each time the multivariate linear regression prediction model is validated, a subset is selected as a historical validation set;

[0040] After the multivariate linear regression prediction model is trained, a historical validation set from the k subsets is selected to evaluate the multivariate linear regression prediction model, and the process is repeated k times, each time a different historical validation set is selected from the k subsets to verify and evaluate the multivariate linear regression prediction model;

[0041] The mean square error, root mean square error and coefficient of determination were calculated through the historical validation set to verify and evaluate the multivariate linear regression prediction model;

[0042] Until all k subsets have completed the verification and evaluation of the multiple linear regression prediction model, the verified multiple linear regression prediction model is obtained.

[0043] Preferably, the expression of the mean square error is:

[0044]

[0045] In the formula, n is the number of diabetic patients;

[0046] is the actual blood glucose value of the i-th diabetic patient;

[0047] y i The predicted blood glucose value for the i-th diabetic patient;

[0048] is the prediction error;

[0049] The expression of the root mean square error is:

[0050]

[0051] In the formula, n is the number of diabetic patients;

[0052] is the actual blood glucose value of the i-th diabetic patient;

[0053] y i The predicted blood glucose value for the i-th diabetic patient;

[0054] is the prediction error;

[0055] The expression of the determination coefficient is:

[0056]

[0057] In the formula, This is the true blood sugar level for diabetic patients.

[0058] Accordingly, the present invention also proposes an artificial intelligence-processed blood glucose data prediction and processing system for diabetic patients, comprising: an acquisition module; a processing module; a prediction module;

[0059] The acquisition module is used to obtain historical data of diabetic patients;

[0060] The processing module is used to perform data preprocessing on the historical data to obtain a preprocessed historical data set;

[0061] The prediction module is used to construct a multiple linear regression prediction model through the historical data set; and predict the blood sugar level of the diabetic patient based on the multiple linear regression prediction model.

[0062] Compared with the prior art, the embodiments of the present invention have at least the following technical advantages:

[0063] From the analysis of the above-mentioned method for predicting and processing blood glucose data of diabetic patients provided by the present invention, it can be known that in specific applications, historical data of diabetic patients are obtained, and the historical data include historical blood glucose values ​​and physiological indicator data; the historical blood glucose values ​​of diabetic patients and the physiological indicator data that affect the blood glucose values ​​are preprocessed by removing abnormal values ​​and standardizing, so that the historical data can avoid some data affecting the prediction ability of the model during the training model, and the model will not be disturbed by these erroneous data, thereby improving the prediction ability of normal situations.

[0064] Furthermore, using the multiple linear regression model as a framework, the historical blood glucose values ​​and physiological index data in the historical data set are used as input independent variables for regression analysis. In the regression analysis, a feature matrix is ​​constructed for the independent variables, and a set of features in the feature matrix is ​​used as input target data to predict the blood glucose value and obtain the regression coefficient; the regression coefficient is then evaluated by the least square method to minimize the sum of squares of the error between the predicted blood glucose value and the actual blood glucose value, and the estimated regression coefficient is obtained. The final multiple linear regression prediction model is constructed by the regression coefficient;

[0065] Furthermore, the historical data set is evenly divided into k subsets by using k-fold cross validation, and one of the subsets is selected as the historical validation set. After the multivariate linear regression prediction model is trained by the remaining subsets, the multivariate linear regression prediction model is evaluated by the historical validation set; and at the same time, the mean square error, root mean square error and determination coefficient are calculated by the historical validation set to verify the indicators of the multivariate linear regression prediction model, which can improve the accuracy of the prediction and reduce the error of the prediction; repeat k times until all subsets have evaluated the multivariate linear regression prediction model;

[0066] Furthermore, new real-time data of diabetic patients are collected and acquired, and the newly acquired real-time data of diabetic patients are predicted using a multivariate linear regression prediction model to obtain the blood glucose values ​​of diabetic patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0068] Figure 1 This is an overall flow chart of a method for predicting and processing blood sugar data of a diabetic patient provided in Embodiment 1 of the present invention;

[0069] Figure 2A method for predicting and processing blood glucose data of a diabetic patient provided in the first embodiment of the present invention is to construct a multivariate linear regression prediction model and predict blood glucose values;

[0070] Figure 3 A schematic diagram of a preliminary predicted blood sugar value of a method for predicting and processing blood sugar data of a diabetic patient provided in Embodiment 1 of the present invention;

[0071] Figure 4 A multivariate linear regression prediction model verification flow chart of a method for predicting and processing blood glucose data of a diabetic patient provided in Example 1 of the present invention;

[0072] Figure 5 A flowchart of a system for predicting and processing blood glucose data of a diabetic patient using artificial intelligence provided in Embodiment 2 of the present invention;

[0073] Figure 6 A storage medium according to the third embodiment of the present invention;

[0074] Label: acquisition module 10, processing module 20, prediction module 30; processor 1110; communication interface 1120; memory 1130; computer storage medium 1140. DETAILED DESCRIPTION

[0075] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] The present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings.

[0077] Embodiment 1

[0078] like Figure 1 As shown, the present invention proposes a method for predicting and processing blood sugar data of a diabetic patient, comprising the following steps:

[0079] S1: Obtain historical data of diabetic patients;

[0080] The historical data includes: historical blood sugar values ​​and physiological index data;

[0081] It should be noted that historical data of diabetic patients were collected, including historical blood sugar values ​​and physiological index data; at the same time, physiological index data included age, disease course, diabetes classification, myocardial infarction / cerebral infarction / renal failure / diabetic foot, BMI, DPP-4, GLP-1, SGLT-2, SGLT-3, α-glucosidase inhibitors (voglibose tablets), α-glucosidase inhibitors (acarbose capsules), non-SU insulin secretagogue nateglinide tablets, non-SU insulin secretagogue repaglinide, glinides, sulfonylureas, aldose reduction, proenzyme inhibitors, insulin sensitizers, biguanides, premixes, premixes (i u), premixed intravenous drip, long-acting, long-acting IU, a total of 24 items (that is, the 24 feature information is used as the main reference data); and not limited to historical blood sugar values ​​(including blood sugar values ​​1 hour, 2 hours, and 3 hours before meals), pre-meal and post-meal markers, carbohydrate intake, protein intake, fat intake, exercise duration, exercise intensity index, body mass index, systolic blood pressure, diastolic blood pressure, heart rate, body temperature, insulin injection dose, etc. (that is, the features not limited to are not the main reference data), which may affect the subsequent construction of the multivariate linear regression prediction model for the analysis and prediction of the independent variables, thereby affecting the results of the dependent variables and causing the model to have identification deviations;

[0082] S2: performing data preprocessing on the historical data to obtain a preprocessed historical data set;

[0083] It should be noted that the collected historical data are preprocessed, including preprocessing such as removing missing values, standardization, and removing outliers, to obtain a historical data set; this allows the historical data set to delete or fill in missing data, so that the model can better utilize all available data and avoid affecting the reliability of the training process or results due to missing values; if there are missing values ​​in the blood glucose data, filling in the missing values ​​(such as using the mean, median, or interpolation by other methods) can ensure that the model will not "skip" these useful information due to missing data; at the same time, if extremely high blood glucose values ​​appear in some data records (such as values ​​exceeding the normal range), these data may be input errors; by processing some outliers, the model will not be disturbed by these erroneous data, thereby improving the prediction ability for normal conditions;

[0084] S3: constructing a multiple linear regression prediction model through the historical data set; and predicting the blood glucose level of the diabetic patient based on the multiple linear regression prediction model.

[0085] It should be noted that the multiple linear regression prediction model is constructed by using the historical data set as a framework of the multiple linear regression model; the multiple linear regression prediction model is more prominent in predictive ability and has improved accuracy, and can predict blood sugar values ​​based on the newly collected data of diabetic patients, which is more accurate than the basic multiple linear regression model;

[0086] Specifically, Figure 2 As shown, in step S3, a multiple linear regression prediction model is constructed through the historical data set; the blood glucose level of the diabetic patient is predicted based on the multiple linear regression prediction model, and the specific operation steps are as follows:

[0087] S31: using a multiple linear regression model to perform regression analysis on the historical blood glucose values ​​and physiological index data in the historical data set as input independent variables to form a feature matrix X;

[0088] Each row of the feature matrix represents a data sample, and each column represents a feature;

[0089] S32: a set of features in the feature matrix: x1, x2, ..., x n ; Use the input target data to predict the blood sugar value y, the expression is:

[0090] y=β0+β1·X1+β2·X2+…+β n ·X n +∈;

[0091] Where y is the predicted blood glucose value (i.e., the dependent variable, the predicted blood glucose value of diabetic patients);

[0092] β0 is the intercept term;

[0093] β0,β1,…,β n ; is the regression coefficient (i.e., the parameter to be estimated, and the regression coefficient of the number of words is evaluated and the multivariate linear regression prediction model is constructed in step S33);

[0094] x1,x2,…,x n is the input feature (i.e., the independent variable, x1 is the age of the diabetic patient, x2 is the course of disease, x3 is the type of diabetes, and so on, the features in the historical blood sugar values ​​and physiological index data);

[0095] ∈ is the error term;

[0096] It should be noted that the historical blood sugar value and the characteristics in the physiological index data (ie, the age, course of disease, diabetes type and other 24 information in the explanation of step S1) are used to make a preliminary prediction result (ie, Figure 3 The historical blood glucose values ​​(including the blood glucose values ​​of the previous 1 hour, 2 hours, and 3 hours), pre-meal and post-meal markings, etc.) are used as independent variables to predict the known information of the dependent variable. These independent variables may be different dimensions or features and can be represented by numbers, categorical variables, etc.

[0097] In regression analysis, these independent variables are used to construct a feature matrix, where each row represents a sample (a data point) and each column represents an independent variable (feature). The feature matrix is ​​a collection of all independent variable data. For example, there are 100 diabetic patients as samples, and there are historical data of 100 diabetic patients. Each diabetic patient has features in historical blood sugar values ​​and physiological index data (i.e., the features in the explanation of step S1, the number of features is n). Then the feature matrix X will form a matrix of 100 rows and n columns:

[0098]

[0099] S33: Using the historical data set to calculate the regression coefficients (i.e., β0, β1, ..., β n ) to estimate:

[0100]

[0101] Where X is the feature matrix;

[0102] X T is the transpose of the feature matrix X;

[0103] (X T X) -1 are feature matrices X and X T The inverse matrix of multiplication;

[0104] y is the predicted blood glucose value;

[0105] is the estimated value of the regression coefficient;

[0106] The regression coefficients are obtained by fitting the multivariate linear regression model through the historical data set. Thus, a multiple linear regression prediction model is obtained;

[0107] The basic form of the multivariate linear regression prediction model is:

[0108]

[0109] In the formula, is the real blood sugar value;

[0110] y is the predicted blood glucose value;

[0111] X is the feature matrix, which contains all the independent variables;

[0112] β is the regression coefficient, which is the parameter that needs to be estimated in regression analysis and is usually obtained through data fitting;

[0113] ∈ is the error term, which represents the difference between the model prediction value and the true value;

[0114] It should be noted that the regression coefficient is the core of the multivariate linear regression prediction model, which determines the degree of influence of each independent variable (feature) on the dependent variable; therefore, the regression coefficient needs to be estimated, and the goal of the multivariate linear regression prediction model is to estimate the regression coefficient β, which describes the weight or influence of each feature in predicting the dependent variable; the regression coefficient is estimated by the Ordinary Least Squares (OLS), and the goal of the least squares method is to find a set of regression coefficients β that minimizes the sum of squares of the error (residual) between the predicted value and the true value of the model; through the estimated regression coefficient β^, a multivariate linear regression prediction model can be constructed, and the multivariate linear regression prediction model is used to predict the data of diabetic patients to obtain new blood glucose values;

[0115] S34: Optimizing the multivariate linear regression prediction model by cross-validation to obtain a final multivariate linear regression prediction model;

[0116] It should be noted that the multivariate linear regression prediction model can be verified through cross-validation to prevent the problem of overfitting; at the same time, the regression prediction analysis of the multivariate linear regression prediction model is performed by calculating the mean square error, root mean square error, and determination coefficient, so as to achieve the purpose of optimizing the multivariate linear regression prediction model; the difference between the predicted value and the actual value of the multivariate linear regression prediction model is measured by the mean square error, which is a performance evaluation indicator commonly used in regression analysis; the calculation of the root mean square error can represent the standard deviation of the prediction error of the multivariate linear regression prediction model, which can reflect the error size of the prediction of the multivariate linear regression prediction model; the calculation of the determination coefficient is also called the goodness of fit, which measures the ability of the multivariate linear regression prediction model to explain the variability of the data, and reflects how much proportion of the real data changes can be explained by the multivariate linear regression prediction model;

[0117] S35: Acquire the latest real-time data of the diabetic patient; predict the latest real-time data of the diabetic patient based on the multivariate linear regression prediction model to obtain the blood glucose value of the diabetic patient;

[0118] It should be noted that the latest real-time data of the diabetic patients is obtained, and the latest real-time data is input into the trained multivariate linear regression prediction model to output the blood glucose value of the diabetic patients;

[0119] Specifically, Figure 4 As shown, in step S34, the multivariate linear regression prediction model is optimized and predicted by cross validation to obtain the final multivariate linear regression prediction model. The specific operation steps are as follows:

[0120] S341: using k-fold cross validation to evenly divide the historical data set into k subsets, and each time the multivariate linear regression prediction model is validated, a subset is selected as a historical validation set;

[0121] After the multivariate linear regression prediction model is trained, a historical validation set from the k subsets is selected to evaluate the multivariate linear regression prediction model, and the process is repeated k times, each time a different historical validation set is selected from the k subsets to verify and evaluate the multivariate linear regression prediction model;

[0122] It should be noted that the k-fold cross-validation method of cross-validation divides the historical data set into k subsets (such as 10 subsets) on average; each time one of the subsets is selected as the historical validation set, and the remaining k-1 subsets are used as training sets; in each round, the model is trained using the training set, and then the performance of the model is evaluated on the historical validation set (by calculating indicators such as loss, accuracy, and mean square error); repeat k times: this process is repeated k times, and a different subset is selected as the historical validation set each time until all subsets are used as a historical validation set; then the mean square error, root mean square error, and determination coefficient are calculated for each selected historical validation set to verify the evaluation ability of the multivariate linear regression prediction model, thereby evaluating the performance of the multivariate linear regression prediction model;

[0123] S342: Calculate mean square error, root mean square error and coefficient of determination to verify and evaluate the multivariate linear regression prediction model through historical validation sets;

[0124] The expression of the mean square error is:

[0125]

[0126] In the formula, n is the number of diabetic patients;

[0127] is the actual blood glucose value of the i-th diabetic patient;

[0128] y i The predicted blood glucose value for the i-th diabetic patient;

[0129] is the prediction error;

[0130] The expression of the root mean square error is:

[0131]

[0132] In the formula, n is the number of diabetic patients;

[0133] is the actual blood glucose value of the i-th diabetic patient;

[0134] yi The predicted blood glucose value for the i-th diabetic patient;

[0135] is the prediction error;

[0136] The expression of the determination coefficient is:

[0137]

[0138] In the formula, is the actual blood sugar value (i.e. average value) of diabetic patients;

[0139] It should be noted that the mean square error, root mean square error and determination coefficient were calculated through the historical validation set to verify and evaluate the multivariate linear regression prediction model;

[0140] First, Mean Squared Error (MSE) is a commonly used performance evaluation indicator in regression analysis; it measures the difference between the predicted blood glucose value of the multivariate linear regression prediction model and the true blood glucose value; the mean squared error calculates the difference between the predicted blood glucose value of the multivariate linear regression prediction model and the true blood glucose value. The smaller the difference, the smaller the mean squared error, indicating that the prediction accuracy of the model is higher; the mean squared error is the square of the error, so it has a stronger penalty effect on larger errors. Large errors will be magnified after being squared, making the model pay more attention to reducing larger deviations;

[0141] The root mean square error (RMSE) is the square root of the MSE, which represents the standard deviation of the prediction error of the multivariate linear regression prediction model. The root mean square error can be regarded as the standard deviation of the prediction error of the multivariate linear regression prediction model, reflecting the typical size of the error of the multivariate linear regression prediction model. The smaller the value, the more accurate the prediction result. The root mean square error is the square root of the mean square error, which provides the scale of the error, and the unit is the same as the unit of the data.

[0142] The coefficient of determination indicates the degree of fit of the multiple linear regression prediction model to the data. The closer it is to 1, the better the multiple linear regression prediction model is. The closer it is to 0, the better the multiple linear regression prediction model is. It measures the ability of the multiple linear regression prediction model to explain the variability of the data. The closer it is to 1, the better the multiple linear regression prediction model is.

[0143] S343: until all k subsets have completed the verification and evaluation of the multiple linear regression prediction model, and the verified multiple linear regression prediction model is obtained;

[0144] It should be noted that by evaluating and verifying the mean square error, root mean square error and determination coefficient of the multivariate linear regression prediction model through each historical verification set, the prediction ability of the multivariate linear regression prediction model can be understood; and until all k subsets have completed the verification and evaluation of the multivariate linear regression prediction model, a complete prediction ability evaluation of the multivariate linear regression prediction model can be performed;

[0145] Embodiment 2

[0146] like Figure 5 As shown, accordingly, the present invention also proposes an artificial intelligence-processed blood glucose data prediction and processing system for diabetic patients, comprising: an acquisition module 10; a processing module 20; a prediction module 30;

[0147] The acquisition module 10 is used to acquire historical data of diabetic patients;

[0148] The processing module 20 is used to perform data preprocessing on the historical data to obtain a preprocessed historical data set;

[0149] The prediction module 30 is used to construct a multiple linear regression prediction model through the historical data set; and predict the blood glucose level of the diabetic patient based on the multiple linear regression prediction model.

[0150] like Figure 6 As shown, on the other hand, this embodiment 3 is based on the method for predicting and processing blood glucose data of a diabetic patient provided in the embodiment 1 of the invention, and further provides a computer storage medium 1140 (hereinafter referred to as storage medium). It is a schematic diagram of the computer storage medium structure framework provided in the embodiment 3 of the present invention, which includes:

[0151] A memory 1130, used for storing computer programs;

[0152] The communication interface 1120 is used to connect the memory 1130 to the processor 1110;

[0153] The processor 1110 is used to execute a computer program to implement a method for predicting and processing blood glucose data of a diabetic patient involved in Example 1 disclosed in combination with any of the above-mentioned implementation modes.

[0154] In summary, the present invention provides a method and system for predicting and processing blood glucose data of a diabetic patient, which obtains historical data of the diabetic patient, and the historical data includes historical blood glucose values ​​and physiological index data; performs preprocessing such as removing abnormal values ​​and standardizing the historical blood glucose values ​​of the diabetic patient and the physiological index data that affect the blood glucose values, so that the historical data can avoid some data from affecting the prediction ability of the model during the training process of the model, and the model will not be disturbed by these erroneous data, thereby improving the prediction ability of normal situations;

[0155] Furthermore, using the multiple linear regression model as a framework, the historical blood glucose values ​​and physiological index data in the historical data set are used as input independent variables for regression analysis. In the regression analysis, a feature matrix is ​​constructed for the independent variables, and a set of features in the feature matrix is ​​used as input target data to predict the blood glucose value and obtain the regression coefficient; the regression coefficient is then evaluated by the least square method to minimize the sum of squares of the error between the predicted blood glucose value and the actual blood glucose value, and the estimated regression coefficient is obtained. The final multiple linear regression prediction model is constructed by the regression coefficient;

[0156] Furthermore, the historical data set is evenly divided into k subsets by using k-fold cross validation, and one of the subsets is selected as the historical validation set. After the multivariate linear regression prediction model is trained by the remaining subsets, the multivariate linear regression prediction model is evaluated by the historical validation set; and at the same time, the mean square error, root mean square error and determination coefficient are calculated by the historical validation set to verify the indicators of the multivariate linear regression prediction model, which can improve the accuracy of the prediction and reduce the error of the prediction; repeat k times until all subsets have evaluated the multivariate linear regression prediction model;

[0157] Furthermore, new real-time data of diabetic patients are collected and acquired, and the newly acquired real-time data of diabetic patients are predicted using a multivariate linear regression prediction model to obtain the blood glucose values ​​of diabetic patients.

[0158] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. A person skilled in the art may modify the technical solutions described in the above embodiments, or replace part or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting and processing blood sugar data of a diabetic patient, characterized in that: The steps are as follows: Obtain historical data of diabetic patients; Performing data preprocessing on the historical data to obtain a preprocessed historical data set; A multiple linear regression prediction model is constructed using the historical data set; and the blood glucose level of the diabetic patient is predicted based on the multiple linear regression prediction model.

2. The method for predicting and processing blood sugar data of a diabetic patient according to claim 1, characterized in that: The historical data includes: historical blood sugar values ​​and physiological index data.

3. The method for predicting and processing blood sugar data of a diabetic patient according to claim 2, characterized in that: The multivariate linear regression prediction model is constructed by using the historical data set. The specific steps are as follows: Using a multiple linear regression model, the historical blood sugar values ​​and physiological index data in the historical data set are used as input independent variables for regression analysis to form a feature matrix X; Each row of the feature matrix represents a data sample, and each column represents a feature; A set of features in the feature matrix: x1, x2,…, x n ; Use the input target data to predict the blood sugar value y, the expression is: y=β0+β1·X1+β2·X2+…+β n ·X n +∈; Where y is the predicted blood glucose value; β0 is the intercept term; β0,β1,…,β n ; is the regression coefficient; x1,x2,…,x n is the input feature; ∈ is the error term; The regression coefficients are estimated using the least squares method using the historical data set: Where X is the feature matrix; X T is the transpose of the feature matrix X; X T X)- 1 are feature matrices X and X T The inverse matrix of multiplication; y is the predicted blood glucose value; is the estimated value of the regression coefficient; The regression coefficients are obtained by fitting the multivariate linear regression model through the historical data set. Thus, a multiple linear regression prediction model is obtained; The multiple linear regression prediction model is optimized and predicted by cross-validation to obtain the final multiple linear regression prediction model.

4. The method for predicting and processing blood sugar data of a diabetic patient according to claim 3, characterized in that: The basic form of the multivariate linear regression prediction model is: In the formula, is the real blood sugar value; y is the predicted blood glucose value; X is the feature matrix, which contains all the independent variables; β is the regression coefficient, which is the parameter that needs to be estimated in regression analysis and is usually obtained through data fitting; ∈ is the error term, which represents the difference between the model prediction value and the true value.

5. A method for predicting and processing blood sugar data of a diabetic patient according to claim 4, characterized in that: Based on the multivariate linear regression prediction model, the blood glucose level of the diabetic patient is predicted to obtain the latest real-time data of the diabetic patient; based on the multivariate linear regression prediction model, the latest real-time data of the diabetic patient is predicted to obtain the blood glucose level of the diabetic patient.

6. A method for predicting and processing blood sugar data of a diabetic patient according to claim 5, characterized in that: The multivariate linear regression prediction model is optimized and predicted by cross validation to obtain the final multivariate linear regression prediction model. The specific operation steps are as follows: The historical data set is evenly divided into k subsets by using k-fold cross validation, and each time the multivariate linear regression prediction model is validated, a subset is selected as a historical validation set; After the multivariate linear regression prediction model is trained, a historical validation set from the k subsets is selected to evaluate the multivariate linear regression prediction model, and the process is repeated k times, each time a different historical validation set is selected from the k subsets to verify and evaluate the multivariate linear regression prediction model; The mean square error, root mean square error and coefficient of determination were calculated through the historical validation set to verify and evaluate the multivariate linear regression prediction model; Until all k subsets have completed the verification and evaluation of the multiple linear regression prediction model, the verified multiple linear regression prediction model is obtained.

7. A method for predicting and processing blood sugar data of a diabetic patient according to claim 6, characterized in that: The expression of the mean square error is: In the formula, n is the number of diabetic patients; is the actual blood glucose value of the i-th diabetic patient; y i The predicted blood glucose value for the i-th diabetic patient; -y i is the prediction error; The expression of the root mean square error is: In the formula, n is the number of diabetic patients; is the actual blood glucose value of the i-th diabetic patient; y i The predicted blood glucose value for the i-th diabetic patient; -y i is the prediction error; The expression of the determination coefficient is: In the formula, This is the actual blood sugar level for diabetic patients.

8. An artificial intelligence-processed blood sugar data prediction and processing system for diabetic patients, characterized in that: include: Acquisition module; Processing module; Prediction module; The acquisition module is used to obtain historical data of diabetic patients; The processing module is used to perform data preprocessing on the historical data to obtain a preprocessed historical data set; The prediction module is used to construct a multiple linear regression prediction model through the historical data set; and predict the blood sugar level of the diabetic patient based on the multiple linear regression prediction model.

Images