Method and application for evaluating postprandial blood glucose response of human body based on continuous blood glucose data

Abstract

The method and application of the present invention for evaluating human postprandial blood glucose response based on continuous blood glucose data obtain continuous blood glucose test values and meal start time; match the first blood glucose recording point after eating according to the meal time and intercept the blood glucose data 2 hours after the meal and preprocess the data; apply the continuous blood glucose data obtained by FPCA to construct four postprandial blood glucose response patterns based on the top two characteristic functions ranked by explanatory power obtained by the FPCA; and by comparing the postprandial blood glucose response mean curves of the four blood glucose response patterns, it is proved that these four response patterns can comprehensively evaluate postprandial blood glucose from two aspects of blood glucose concentration range and whether fasting steady state can be quickly restored.

Description

Technical Field

[0001] The present invention belongs to the technical field of nutritional intervention, and in particular relates to a method and application for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data. Background Art

[0002] In recent years, the importance of postprandial blood glucose management in metabolic health and diabetes control has garnered widespread attention. Studies have shown that significant increases in postprandial blood glucose are a significant risk factor for metabolic disorders such as prediabetes, type 2 diabetes, and cardiovascular disease. Effectively controlling postprandial blood glucose fluctuations not only improves insulin resistance and overall pancreatic function but also helps prevent diabetic complications. With scientific advances, numerous medications and therapies have been designed to reduce postprandial blood glucose fluctuations.

[0003] Fluctuations in blood sugar levels after eating vary significantly between individuals and are affected by multiple factors, including genes, intestinal microbiota, lifestyle, metabolic status, and the composition and combination of food. Traditional standardized blood sugar management methods often ignore these individual differences, resulting in limited effectiveness in controlling postprandial blood sugar fluctuations. To address this challenge, researchers have gradually turned to personalized intervention strategies, combining multidimensional biological data to develop personalized diet plans, nutritional interventions, and treatment plans to more effectively manage postprandial blood sugar. This precise intervention strategy based on individual characteristics shows great potential in improving metabolic health, reducing the risk of diabetes and its complications, and has broad application prospects.

[0004] Current methods for assessing postprandial glycemic response primarily rely on single-point blood glucose values, such as peak values and postprandial glucose increment area, to characterize an individual's postprandial glycemic response to various nutritional interventions and inform dietary choices. However, these two metrics have limitations in clinical practice. First, there is no clear clinical reference range for postprandial glucose increment area, making it difficult to apply to clinical glucose management across diverse populations. Furthermore, insulin is crucial for postprandial glucose regulation, and the first and second phases of insulin release have distinct physiological significance. Furthermore, the proper functioning of these two releases directly impacts an individual's glucose metabolism. However, postprandial glucose increment area fails to identify the dynamic mechanisms by which insulin regulates postprandial glucose. Furthermore, while a wealth of research supports the importance of single-point blood glucose values, such as peak values, one-hour postprandial glucose values, and two-hour postprandial glucose values, in the development and progression of diabetes, these values cannot fully capture the characteristics of the postprandial glucose response in terms of shape and trend. These values provide only limited information when assessing postprandial glucose response. Given the complexity and dynamic nature of human physiology, identifying postprandial glucose characteristics over time, compared to single-point glucose values or overall metrics, can further demonstrate the body's ability to regulate glucose metabolism.

[0005] Therefore, how to solve the problem that a single-point blood glucose value cannot evaluate the changing characteristics of continuous blood glucose data, and the blood glucose increment area commonly used in continuous blood glucose data cannot evaluate the postprandial blood glucose control range and quickly obtain fasting steady state of postprandial blood glucose, provide individuals with more accurate postprandial blood glucose response pattern assessment, and open up a new path to achieve personalized postprandial blood glucose response assessment, is a technical problem that needs to be urgently solved by technical personnel in this field. Summary of the Invention

[0006] The first purpose of the present invention is to comprehensively evaluate and classify postprandial blood glucose response patterns from the time dimension. In response to the problem that the existing technology cannot comprehensively describe postprandial blood glucose response from the time dimension, a postprandial blood glucose evaluation and classification model based on a time series algorithm is provided.

[0007] To this end, the above-mentioned purpose of the present invention is achieved through the following technical solutions:

[0008] A method for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data, characterized in that it comprises the following steps:

[0009] S1, data collection: The collected sample data include meal time points, continuous blood glucose concentrations and their corresponding time points;

[0010] S2, data preprocessing: matching meal time points, intercepting blood glucose data, processing missing values to a single post-meal response blood glucose data point of no less than 9 and standardization processing: calculating the mean μ and standard deviation σ of the post-meal blood glucose data set, using the formula The collected data were converted into a standardized data set with a mean of 0 and a standard deviation of 1 to eliminate dimensional differences and obtain a postprandial blood glucose data set;

[0011] S3, perform FPCA processing on the postprandial blood glucose dataset in step S2: smooth the postprandial blood glucose dataset using a Gaussian kernel function, process the boundary points by mirror filling, and calculate the weighted average of each data point to reduce noise interference and preserve the data trend; then, construct a covariance matrix based on the smoothed postprandial blood glucose dataset, and calculate the eigenvalues and eigenvectors; finally, sort the eigenvalues by size, and select the eigenvectors corresponding to the first two largest eigenvalues (the top two in contribution) as eigenfunction I and eigenfunction II;

[0012] S4, calculate the projection value of the postprandial blood glucose curve on characteristic function I and characteristic function II by inner product: Based on step S3, calculate the projection value of the postprandial blood glucose curve of each meal on characteristic function I and characteristic function II according to characteristic function I and characteristic function II of the postprandial blood glucose dataset.

[0013] S5, respectively calculating the median value of the projection value of characteristic function I and the median value of the projection value on characteristic function II in the postprandial blood glucose dataset;

[0014] S6, Postprandial blood glucose response assessment: By calculating the projection value of the blood glucose response of a single meal on characteristic function I and the projection value on characteristic function II, they are compared with the median value of the projection value of characteristic function I and the median value of the projection value of characteristic function II of the postprandial blood glucose dataset to assess the postprandial blood glucose response.

[0015] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions:

[0016] As a preferred technical solution of the present invention: in step S1, the collected data includes recording the meal time of each meal, and obtaining continuous blood glucose concentration and its corresponding time point after the meal through CGM or venous blood samples, wherein the blood glucose monitoring time after each meal is greater than or equal to 2 hours, and the monitoring interval is the same, and the interval between two time points does not exceed 15 minutes.

[0017] As a preferred technical solution of the present invention: in step S2, the first blood glucose concentration value recording point after the meal is matched according to each meal time point of the sample and the blood glucose data 2 hours after the meal is intercepted, and the blood glucose data point after each meal is made into a list and stored in the post-meal blood glucose data set.

[0018] As a preferred technical solution of the present invention: Step S3 includes the following steps:

[0019] S3.1 Data smoothing: Use the Gaussian kernel function parameters to smooth the postprandial blood glucose dataset, apply mirror filling to the boundary points, traverse each data point, calculate its weighted average, and finally output the smoothed postprandial blood glucose dataset to reduce noise interference and preserve data trends;

[0020] S3.2 Calculate the covariance matrix: Calculate the mean of all samples at each time point. Two time points i and j constitute a pair of time points. Use a double loop to traverse each pair of time points (i, j) to calculate the covariance. For each pair of time points, calculate the covariance of all samples and fill the calculated covariance values into the covariance matrix to obtain the covariance matrix of the postprandial blood glucose dataset.

[0021] S3.3 Eigenfunction decomposition: Decompose the covariance matrix to obtain eigenvalues and eigenvectors. The obtained eigenvectors represent the eigenfunction;

[0022] S3.4 Eigenfunction selection: Sort the eigenvalues according to their size, and select the eigenvectors corresponding to the first two largest eigenvalues (the top two in contribution) as eigenfunction I and eigenfunction II.

[0023] As a preferred technical solution of the present invention: in step S3.1, the smoothing parameter is selected as 0.1.

[0024] As a preferred technical solution of the present invention: in step S3.2, the smoothed postprandial blood glucose dataset is X, which contains n samples and m time points. The covariance matrix C is an m×m matrix, in which the element in the i-th row and j-th column represents the covariance between the i-th time point and the j-th time point. The covariance calculation formula is

[0025] in, refers to the blood glucose concentration value of the kth sample at the i-th time point, Refers to the blood glucose concentration value of the kth sample at the jth time point, is the mean value at the i-th time point, is the mean at the jth time point, and n is the number of samples.

[0026] As a preferred technical solution of the present invention: in step S3.4, the cumulative explained variance of characteristic function I and characteristic function II is greater than 80%, and it is determined that the first two characteristic functions obtained by FPCA can fully characterize the variability of postprandial blood glucose response while reducing the dimension of the features;

[0027] The explained variance of each eigenvalue λi is calculated by the following formula:

[0028] Proportion of variance explained = ,

[0029] Where: λi is the i-th eigenvalue;

[0030] is the sum of all eigenvalues.

[0031] As a preferred technical solution of the present invention: in step S6, four different postprandial blood glucose response patterns are constructed based on the median of the projection values of the first two characteristic functions of the postprandial blood glucose data set, wherein the projection value of the single blood glucose response data on the characteristic function I and the characteristic function II is less than the median and is judged as L, and greater than or equal to the median and is judged as H:

[0032] HH mode: the characteristic function I is estimated to be H, and the characteristic function II is estimated to be H;

[0033] HL mode: the evaluation of characteristic function I is H, and the evaluation of characteristic function II is L;

[0034] LH mode: the evaluation of characteristic function I is judged as L, and the evaluation of characteristic function II is judged as H;

[0035] LL mode: The valuation of characteristic function I is judged to be L, and the valuation of characteristic function II is judged to be L.

[0036] The second object of the present invention is to provide an application of the method for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data.

[0037] To this end, the above-mentioned purpose of the present invention is achieved through the following technical solutions:

[0038] The application of the method for evaluating the postprandial blood glucose response of human body based on continuous blood glucose data is characterized in that

[0039] It is used to judge whether the postprandial blood sugar response is good or not. Among them, the HL pattern is judged as a good response pattern; the LH pattern, HH pattern and LL pattern are judged as poor response patterns.

[0040] Compared with the prior art, the method and application of the present invention for evaluating human postprandial blood glucose response based on continuous blood glucose data have the following beneficial effects: the present invention uses functional principal component analysis to extract characteristic function I and characteristic function II from the postprandial blood glucose data set to capture the main trends and variations of the postprandial blood glucose curve, effectively filter out noise and non-important information, and retain the main change patterns of the data; by calculating the projection values of the postprandial blood glucose curve on characteristic function I and characteristic function II, the projection values reflect the similarity between the postprandial blood glucose curve and the characteristic function, quantify the contribution of each curve to the main change pattern, and facilitate the quantification of individual blood glucose response characteristics; using the median of the projection values of characteristic function I and characteristic function II in the postprandial blood glucose data set as a reference benchmark, the projection value of a single meal is compared with the median, the blood glucose response intensity at each time point is evaluated, and the main blood glucose fluctuation pattern is identified; the projection value of the single meal blood glucose response curve on characteristic function I and characteristic function II is calculated to identify the dynamic characteristics of postprandial blood glucose changes, providing a basis for opening up new paths for evaluating personalized human postprandial blood glucose and dynamically adjusting nutritional intervention.

[0041] Specifically, in the present invention, FPCA analysis is used to capture the entire fluctuation of blood glucose within 2 hours after a meal. By capturing the instantaneous changes and overall trends of blood glucose, a complete record of the dynamic changes of blood glucose and the fluctuations at different time points is ensured, solving the problems of inaccuracy and non-comprehensiveness of obtaining the value of a certain time point or a certain comprehensive quantitative indicator alone; by comprehensively analyzing data from multiple time points, a more accurate postprandial blood glucose assessment is provided, especially in terms of the sensitivity and recovery ability of glucose metabolism regulation; in the present invention, FPCA is used to extract the first two characteristic functions, which can explain >80% of the variance of the variability of postprandial blood glucose response, significantly reducing the data dimension while retaining the main variability of the data, solving the problem of overly lengthy and complex data processing in traditional methods, making the analysis of large-scale postprandial blood glucose data more efficient, while still maintaining the explanatory power of data variability, and realizing dimensionality reduction and efficient analysis of postprandial blood glucose data; the application of FPCA can effectively smooth noise data, reduce the impact of monitoring errors on blood glucose value assessment results, and improve the stability and accuracy of the results.

[0042] The present invention provides a method and application for evaluating the postprandial blood glucose response of the human body based on continuous blood glucose data. The method extracts the main trends and changes of the postprandial blood glucose curve through characteristic function I and characteristic function II, combines projection value calculation and median comparison, accurately quantifies individual blood glucose response characteristics, and identifies dynamic change patterns. More comprehensive blood glucose fluctuation information can be obtained, which helps to more accurately evaluate the changing trends and patterns of postprandial blood glucose. By analyzing blood glucose values at multiple time points and integrating information on the postprandial blood glucose control range and rapid recovery to steady state, the postprandial blood glucose response can be better evaluated. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flow chart of a method for assessing postprandial blood glucose response in humans based on continuous blood glucose data;

[0044] Figure 2 Visualization of characteristic function I and characteristic function II and four postprandial blood glucose response patterns. DETAILED DESCRIPTION

[0045] The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

[0046] like Figure 1-Figure 2 As shown, the method of the present invention for evaluating the postprandial blood glucose response of a human body based on continuous blood glucose data comprises the following steps:

[0047] S1, data collection: The collected sample data include meal time points, continuous blood glucose concentrations and their corresponding time points;

[0048] S2, data preprocessing: matching meal time points, intercepting blood glucose data, processing missing values to a single post-meal response blood glucose data point of no less than 9 and standardization processing: calculating the mean μ and standard deviation σ of the post-meal blood glucose data set, using the formula The collected data were converted into a standardized data set with a mean of 0 and a standard deviation of 1 to eliminate dimensional differences and obtain a postprandial blood glucose data set;

[0049] S3, performing FPCA processing on the postprandial blood glucose dataset in step S2: smoothing the postprandial blood glucose dataset using a Gaussian kernel function, processing boundary points by mirror filling, and calculating the weighted average of each data point to reduce noise interference and retain data trends; then, constructing a covariance matrix based on the smoothed postprandial blood glucose dataset, and calculating the eigenvalues and eigenvectors; finally, sorting by eigenvalue size, selecting the eigenvectors corresponding to the first two largest eigenvalues as the top two contributing eigenfunctions I and II;

[0050] S4, calculate the projection value of the postprandial blood glucose curve on characteristic function I and characteristic function II by inner product: Based on step S3, according to characteristic function I and characteristic function II of the postprandial blood glucose data set, calculate the projection value of the postprandial blood glucose curve at each meal time point on characteristic function I and characteristic function II.

[0051] S5, respectively calculating the median value of the projection value of characteristic function I and the median value of the projection value on characteristic function II in the postprandial blood glucose dataset;

[0052] S6, Postprandial blood glucose response assessment: By calculating the projection value of the blood glucose response of a single meal on characteristic function I and the projection value on characteristic function II, they are compared with the median value of the projection value of characteristic function I and the median value of the projection value of characteristic function II of the postprandial blood glucose dataset to assess the postprandial blood glucose response.

[0053] When collecting data:

[0054] Record the time of each meal and obtain continuous blood glucose concentrations and corresponding time points after each meal through CGM or venous blood sampling. The blood glucose monitoring time after each meal must be greater than or equal to 2 hours, and the monitoring interval must be the same, with the interval between two time points not exceeding 15 minutes. In other words, the number of blood glucose monitoring data points after each meal must be greater than or equal to 9. CGM is preferred for data collection.

[0055] Data preprocessing:

[0056] The first blood glucose concentration value recording point after each meal was matched according to each meal time point and the blood glucose data 2 hours after the meal was intercepted. The blood glucose data points after each meal were stored as a list in the postprandial blood glucose dataset. The list of blood glucose data points with missing values or less than 9 data points in the postprandial blood glucose dataset was supplemented. The postprandial blood glucose dataset was standardized and the data was converted to a mean of 0 and a standard deviation of 1 to eliminate the dimensional differences between the data.

[0057] The postprandial blood glucose dataset in step S2 is processed by FPCA, including the following steps:

[0058] S3.1 Data smoothing: Select the Gaussian kernel function parameters (the smoothing parameter is set to 0.1), calculate the weighted average of each data point with it as the center, apply mirror filling to the boundary points, and finally output the smoothed postprandial blood glucose dataset, effectively reducing noise interference and preserving data trends;

[0059] S3.2 Calculate the covariance matrix: Calculate the mean of all samples at each time point. Two time points i and j constitute a pair of time points. Use a double loop to traverse each pair of time points (i, j) to calculate the covariance. For each pair of time points, calculate the covariance of all samples and fill the calculated covariance values into the covariance matrix to obtain the covariance matrix of the postprandial blood glucose dataset.

[0060] S3.3 Eigenfunction decomposition: Decompose the covariance matrix to obtain eigenvalues and eigenvectors. The obtained eigenvectors represent the eigenfunction;

[0061] S3.4 Feature function selection: From the obtained feature functions, select the top two feature functions that contribute the most to the data based on the variance explained by each feature function (the size of the eigenvalue). If the cumulative explained variance of the top two feature functions is greater than 80%, it is considered that FPCA can explain the trend of the data while reducing the dimension of the features.

[0062] S3.5 Output valuation: Calculate the inner product between the postprandial blood glucose curve of each meal and characteristic function I and characteristic function II to obtain the projection values on characteristic function I and characteristic function II, and calculate the median of the projection values of the postprandial continuous blood glucose data set on the first two characteristic functions.

[0063] In step S3.2, the smoothed postprandial blood glucose dataset is X, which contains n samples and m time points. The covariance matrix C is an m×m matrix, where the element in the i-th row and j-th column represents the covariance between the i-th time point and the j-th time point. The covariance calculation formula is

[0064] in, refers to the blood glucose concentration value of the kth sample at the i-th time point, Refers to the blood glucose concentration value of the kth sample at the jth time point, is the mean value at the i-th time point, is the mean at the jth time point, and n is the number of samples.

[0065] In step S3.4, the cumulative explained variance of characteristic function I and characteristic function II is greater than 80%, indicating that the first two characteristic functions obtained by FPCA can fully characterize the variability of postprandial blood glucose response while reducing the dimension of the features;

[0066] The explained variance of each eigenvalue λi is calculated by the following formula:

[0067] Proportion of variance explained = ,

[0068] Where: λi is the i-th eigenvalue;

[0069] is the sum of all eigenvalues.

[0070] Constructing postprandial blood glucose response patterns: Four different postprandial blood glucose response patterns were constructed based on the median of the projection values of the first two eigenfunctions of the postprandial blood glucose dataset. Among them, the projection value of the single blood glucose response data on the eigenfunction I and the eigenfunction II was judged to be L if it was less than the median, and was judged to be H if it was greater than or equal to the median: HH (eigenfunction I was valued as H, and eigenfunction II was valued as H), HL (eigenfunction I was valued as H, and eigenfunction II was valued as L), LH (eigenfunction I was valued as L, and eigenfunction II was valued as H), and LL (eigenfunction I was valued as L, and eigenfunction II was valued as L). Comparing the mean curves of the four postprandial blood glucose response patterns, the HL pattern was a good response pattern if it met the clinical control range of blood glucose and blood glucose returned to fasting steady state 2 hours after the meal. The LH pattern was a poor response pattern if it did not meet the clinical control range of blood glucose and blood glucose could not return to fasting steady state 2 hours after the meal.

[0071] Application examples of postprandial blood glucose response models

[0072] Based on the population with normal fasting blood glucose, a postprandial continuous blood glucose dataset that meets the above requirements was collected and data preprocessing was performed. The above-mentioned FPCA was then applied for analysis, and the top two characteristic functions were obtained through the covariance function. Their cumulative explained variance was 85.2%, so it is believed that FPCA can fully characterize the variability of the data in this dataset. By outputting the valuation of each list in this postprandial continuous blood glucose dataset on the top two characteristic functions, it was calculated that the median valuation of the first characteristic function of the postprandial blood glucose response of the population with normal fasting blood glucose was 0.085, and the median valuation of the second characteristic function was 0.82.

[0073] After collecting continuous blood glucose data after a meal that meets the above requirements and performing data preprocessing, the data is smoothed using the Gaussian kernel smoothing method (the smoothing parameter is selected as 0.1). The data is then inner-producted with the characteristic function using the following formula to obtain its estimated value on each characteristic function.

[0074]

[0075] Where f(t) is the new input data (a list of postprandial blood glucose data points), is the kth characteristic function, is the estimate of the kth eigenfunction.

[0076] Subsequently, referring to the medians of characteristic function I and characteristic function II calculated for the population, the postprandial blood glucose response pattern of the blood glucose response curve of a single meal was obtained based on the medians of the two to evaluate the postprandial blood glucose response of this meal.

[0077] The method of the present invention for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data has the following characteristics:

[0078] Collect accurate meal times and continuous blood glucose data that meet requirements, and perform corresponding data preprocessing.

[0079] By analyzing postprandial blood glucose through FPCA, the purpose of evaluating postprandial blood glucose can be achieved from the time dimension.

[0080] Four postprandial blood glucose patterns were constructed using the first two eigenfunction estimates from FPCA analysis: HH, HL, LL, and LH. These patterns can be used to assess postprandial blood glucose levels based on both the range of blood glucose control and the ability to quickly regain fasting homeostasis.

[0081] Compared with the prior art, the method of the present invention for evaluating the postprandial blood glucose response of a human body based on continuous blood glucose data has the following beneficial effects:

[0082] Evaluate postprandial blood glucose response from a temporal perspective: The present invention can capture the entire fluctuation characteristics of blood glucose within 2 hours after a meal, rather than just the value at a certain point in time or a comprehensive quantitative indicator, thereby more accurately reflecting the instantaneous changes and overall trends of blood glucose, and providing a more accurate assessment of postprandial blood glucose response, especially in terms of sensitivity and recovery ability of glucose metabolism regulation.

[0083] Reduce data dimensionality and improve computational efficiency: By extracting the first two eigenfunctions through FPCA, the data dimensionality is significantly reduced, making the analysis of large-scale postprandial blood glucose data more efficient while still maintaining the interpretability of data variability, thus avoiding the problem of overly lengthy and complex data processing in traditional methods.

[0084] High accuracy and stability: The application of FPCA can effectively smooth noisy data, reduce the impact of individual fluctuations or monitoring errors on blood glucose assessment results, improve the stability and accuracy of the results, and provide a more reliable basis for clinical decision-making.

[0085] The present invention provides a method and application for evaluating postprandial blood glucose response based on continuous blood glucose data. The method comprises: obtaining continuous blood glucose test values and meal start time; matching the first blood glucose recording point after a meal according to the meal time and intercepting blood glucose data 2 hours after the meal and preprocessing the data; applying FPCA to the continuous blood glucose data, and constructing four postprandial blood glucose response patterns based on the top two characteristic functions ranked by explanatory power obtained by the FPCA; and comparing the postprandial blood glucose response mean curves of the four blood glucose response patterns to prove that the four response patterns can comprehensively evaluate the postprandial blood glucose response in terms of both the blood glucose concentration range and the ability to quickly regain fasting steady state.

[0086] The above-mentioned specific implementation methods are used to illustrate the present invention and are only preferred embodiments of the present invention, rather than limiting the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit of the present invention and the scope of protection of the claims shall fall within the scope of protection of the present invention.

Claims

1. A method for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data, characterized in that: The following steps are involved: S1, data collection: The collected sample data include meal time points, continuous blood glucose concentrations and their corresponding time points; S2, data preprocessing: matching meal time points, intercepting blood glucose data, processing missing values to a single post-meal response blood glucose data point of no less than 9, and standardization processing: calculating the mean μ and standard deviation σ of the post-meal blood glucose data set, using the formula The collected data were converted into a standardized data set with a mean of 0 and a standard deviation of 1 to eliminate dimensional differences and obtain a postprandial blood glucose data set; S3, performing a functional principal component analysis on the postprandial blood glucose dataset in step S2: smoothing the postprandial blood glucose dataset using a Gaussian kernel function, processing boundary points by mirror filling, and calculating the weighted average of each data point to reduce noise interference and retain data trends; then, constructing a covariance matrix based on the smoothed postprandial blood glucose dataset, and calculating the eigenvalues and eigenvectors; finally, sorting the eigenvalues by size, and selecting the eigenvectors corresponding to the first two largest eigenvalues as the top two contributing eigenfunctions I and II; S4, calculating the projection value of the postprandial blood glucose curve on the characteristic function I and the characteristic function II by inner product: Based on step S3, according to the characteristic function I and the characteristic function II of the postprandial blood glucose dataset, calculating the projection value of each postprandial blood glucose curve in the dataset on the characteristic function I and the characteristic function II; S5, respectively calculating the median value of the projection value of characteristic function I and the median value of the projection value on characteristic function II in the postprandial blood glucose dataset; S6, postprandial blood glucose response assessment: The projection value of the blood glucose response of a single meal on characteristic function I and characteristic function II is calculated, and the projection value is compared with the median value of the projection value of characteristic function I and the median value of the projection value of characteristic function II of the postprandial blood glucose dataset to assess the postprandial blood glucose response; Step S3 includes the following steps: S3.1 Data smoothing: Use the Gaussian kernel function parameters to smooth the postprandial blood glucose dataset, apply mirror filling to the boundary points, traverse each data point, calculate its weighted average, and finally output the smoothed postprandial blood glucose dataset to reduce noise interference and preserve data trends; S3.2 Calculate the covariance matrix: Calculate the mean of all samples at each time point. Two time points i and j constitute a pair of time points. Use a double loop to traverse each pair of time points (i, j) to calculate the covariance. For each pair of time points, calculate the covariance of all samples and fill the calculated covariance values into the covariance matrix to obtain the covariance matrix of the postprandial blood glucose dataset. S3.3 Eigenfunction decomposition: Decompose the covariance matrix to obtain eigenvalues and eigenvectors. The obtained eigenvectors represent the eigenfunction; S3.4 Feature function selection: Sort the eigenvalues according to their size, and select the eigenvectors corresponding to the first two largest eigenvalues as feature function I and feature function II, as the top two feature functions in terms of contribution.

2. The method for assessing a human's postprandial blood glucose response based on continuous blood glucose data according to claim 1, wherein: In step S1, the collected data includes recording the meal time of each meal, and obtaining the continuous blood glucose concentration and its corresponding time point after the meal through a continuous blood glucose monitor or venous blood sample, wherein the blood glucose monitoring time after each meal is greater than or equal to 2 hours, and the monitoring interval is the same, and the interval between two time points does not exceed 15 minutes.

3. The method for assessing a human's postprandial blood glucose response based on continuous blood glucose data according to claim 1, wherein: In step S2, the first blood glucose concentration value recording point after the meal is matched according to each meal time point of the sample and the blood glucose data 2 hours after the meal is intercepted, and the blood glucose data point after each meal is made into a list and stored in the post-meal blood glucose data set.

4. The method for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data according to claim 1, wherein: In step S3.1, the smoothing parameter is selected as 0.

1.

5. The method for evaluating a human body's postprandial blood glucose response based on continuous blood glucose data according to claim 1, wherein: In step S3.2, the smoothed postprandial blood glucose dataset is X, which contains n samples and m time points. The covariance matrix C is an m×m matrix, where the element in the i-th row and j-th column represents the covariance between the i-th time point and the j-th time point. The covariance calculation formula is , in, Refers to the blood glucose concentration value of the kth sample at the i-th time point, Refers to the blood glucose concentration value of the kth sample at the jth time point, is the mean value at the i-th time point, is the mean at the jth time point, and n is the number of samples.

6. The method for evaluating a human's postprandial blood glucose response based on continuous blood glucose data according to claim 1, wherein: In step S3.4, the cumulative explained variance of characteristic function I and characteristic function II is greater than 80%, indicating that the first two characteristic functions obtained by functional principal component analysis can fully characterize the variability of postprandial blood glucose response while reducing the dimensionality of the features; The explained variance of each eigenvalue λi is calculated by the following formula: Proportion of variance explained = , Where: λi is the i-th eigenvalue; is the sum of all eigenvalues.

7. The method for assessing a human's postprandial blood glucose response based on continuous blood glucose data according to claim 1, wherein: In step S6, four different postprandial blood glucose response patterns are constructed based on the median of the projection values of the first two characteristic functions of the postprandial blood glucose dataset. The projection value of the single blood glucose response data on the characteristic function I and the characteristic function II is less than the median and is judged as L, and is greater than or equal to L and is judged as H: HH mode: the characteristic function I is estimated to be H, and the characteristic function II is estimated to be H; HL mode: the evaluation of characteristic function I is H, and the evaluation of characteristic function II is L; LH mode: the characteristic function I is estimated as L, and the characteristic function II is estimated as H; LL mode: The valuation of characteristic function I is judged to be L, and the valuation of characteristic function II is judged to be L.

8. The use of the method for evaluating human postprandial blood glucose response based on continuous blood glucose data according to claim 7, characterized in that: It is used to judge whether the postprandial blood sugar response is good or not. Among them, the HL pattern is judged as a good response pattern; the LH pattern, HH pattern and LL pattern are judged as poor response patterns.

Images