Inulin nutrition enrichment intervention effect prediction method, electronic equipment and medium
By establishing a machine learning-based inulin nutrition enhancement intervention effect prediction method, using SHAP values to screen diabetic patients suitable for inulin intervention, the lack of a solution to predict the effect of inulin intervention in the prior art is solved, and the effect of accurately screening patients before intervention is achieved.
Patent Information
- Application Number
- CN202411739730.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-02-25
AI Technical Summary
The prior art lacks solutions to predict the effect of inulin nutritional fortification intervention, and it is difficult to screen diabetic patients suitable for inulin intervention before intervention.
By determining the initial variables based on patient information, analyzing and screening training variables, establishing machine learning models, calculating SHAP values, and predicting the effect of inulin nutritional enhancement intervention.
Predicting and screening of diabetic patients suitable for inulin intervention before intervention was achieved, improving the prediction accuracy of intervention effect.
Smart Images
Figure CN119517310A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical technology, and more specifically to a method for predicting the effect of inulin nutritional fortification intervention, electronic equipment and a medium. Background Art
[0002] In recent years, the incidence of type 2 diabetes mellitus (T2DM) has been on the rise. In addition to traditional drug treatment, nutritional therapy has become an important method for the treatment of T2DM. Inulin is a fructan soluble fiber that can promote the proliferation of probiotics bifidobacteria and lactobacilli. It has been widely used in nutritional intervention treatment of T2DM, but there is currently no program to predict the effect of inulin nutritional fortification intervention.
[0003] Therefore, it is necessary to develop a method, electronic equipment and medium for predicting the effect of inulin nutritional fortification intervention.
[0004] The information disclosed in the background technology section of the present invention is only intended to deepen the understanding of the general background technology of the present invention, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art already known to those skilled in the art. Summary of the invention
[0005] The present invention provides an inulin nutritional fortification intervention effect prediction method, electronic equipment and medium, which can predict and screen diabetic patients suitable for inulin intervention before intervention.
[0006] In a first aspect, the present disclosure provides a method for predicting the effect of inulin nutritional fortification intervention, comprising:
[0007] Based on the patient information, determine the initial variables;
[0008] Analyze the initial variables and screen the training variables;
[0009] Establishing a machine learning model, training the machine learning model according to the training variables, and calculating a SHAP value;
[0010] The patient's training variables were input into the trained model to predict the patient's effect on inulin nutritional fortification intervention.
[0011] Preferably, the initial variables include the patient's gender, age, body mass index, history of type 2 diabetes, fasting blood glucose, blood glucose levels 30 minutes, 60 minutes, 120 minutes, and 180 minutes after a meal, the area under the curve of the postprandial blood glucose change curve, glycosylated hemoglobin, total cholesterol, triglycerides, low-density lipoprotein, high-density lipoprotein, and glycosylated hemoglobin after inulin nutritional fortification intervention.
[0012] Preferably, analyzing the initial variables includes:
[0013] The individual analyses of the initial variables were performed using the “CBCgrps” package, and the associations between these variables were analyzed using the “corrplot” package;
[0014] The initial variables were screened according to LASSO regression in the “glmnet” package.
[0015] Preferably, the training variables include glycated hemoglobin, the difference between fasting blood glucose and 120 minutes after a meal, fasting blood glucose, high-density lipoprotein, age, body mass index, and glycated hemoglobin after inulin nutritional fortification intervention.
[0016] Preferably, the machine learning model is built by XGBoost.
[0017] Preferably, establishing a machine learning model, and training the machine learning model according to the training variables comprises:
[0018] Convert the training variables into DMatrix format, set the learning rate, maximum depth, and number of iterations, and obtain the trained model.
[0019] Preferably, the training variables of the patient are input into the trained model, and the effect of the patient's inulin nutritional fortification intervention is predicted to include:
[0020] The SHAP threshold was set, the patient's training variables were input into the trained model, and the SHAP value of the patient was calculated. If the SHAP value was less than the SHAP threshold, the inulin nutritional fortification intervention was ineffective for the patient. If the SHAP value was greater than the SHAP threshold, the inulin nutritional fortification intervention was effective for the patient.
[0021] In a second aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:
[0022] A memory storing executable instructions;
[0023] A processor runs the executable instructions in the memory to implement the inulin nutritional fortification intervention effect prediction method.
[0024] In a third aspect, the embodiments of the present disclosure further provide a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the method for predicting the effect of inulin nutritional fortification intervention is implemented.
[0025] The methods and apparatus of the present invention have other features and advantages that will be apparent from, or will be described in detail in, the accompanying drawings and subsequent detailed descriptions incorporated herein, which together serve to explain the specific principles of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] The above and other objects, features and advantages of the present invention will become more apparent through a more detailed description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present invention.
[0027] Figure 1 A flow chart showing the steps of a method for predicting the effect of inulin nutritional fortification intervention according to an embodiment of the present invention.
[0028] Figure 2a , Figure 2b Schematic diagrams of LASSO regression analysis of a LASSO coefficient path diagram and a cross curve are respectively shown according to an embodiment of the present invention.
[0029] Figure 3a , Figure 3b , Figure 3c , Figure 3d , Figure 3e , Figure 3f Schematic diagrams of a training set ROC curve, a test set ROC curve, a training set calibration curve, a test set calibration curve, a training set decision curve, and a test set decision curve according to an embodiment of the present invention are respectively shown.
[0030] Figure 4 A schematic diagram showing the importance of training variables using SHAP values according to an embodiment of the present invention is shown.
[0031] Figure 5 A schematic diagram of a SHAP variable dependency graph according to an embodiment of the present invention is shown.
[0032] Figure 6 A schematic diagram of SHAP force graphs of two random patients according to an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0033] The preferred embodiments of the present invention will be described in more detail below. Although the preferred embodiments of the present invention are described below, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.
[0034] Figure 1 A flow chart showing the steps of a method for predicting the effect of inulin nutritional fortification intervention according to an embodiment of the present invention.
[0035] like Figure 1 As shown, the method for predicting the effect of inulin nutritional fortification intervention includes:
[0036] Step 101, determining initial variables according to patient information;
[0037] Step 102, analyzing the initial variables and selecting the training variables;
[0038] Step 103, establishing a machine learning model, training the machine learning model according to the training variables, and calculating the SHAP value;
[0039] Step 104, inputting the patient's training variables into the trained model to predict the effect of the inulin nutritional fortification intervention on the patient.
[0040] In one example, the initial variables include the patient's gender, age, body mass index, history of type 2 diabetes, fasting blood glucose, blood glucose values at 30 minutes, 60 minutes, 120 minutes, and 180 minutes after a meal, the area under the curve of the postprandial blood glucose change curve, glycated hemoglobin, total cholesterol, triglycerides, low-density lipoprotein, high-density lipoprotein, and glycated hemoglobin after inulin nutritional fortification intervention.
[0041] In one example, the analysis of the initial variables includes:
[0042] Individual analyses of the initial variables were performed using the “CBCgrps” package, and the associations between these variables were analyzed using the “corrplot” package;
[0043] Initial variables were screened according to LASSO regression in the “glmnet” package.
[0044] In one example, the training variables include glycated hemoglobin, the difference between fasting and 120 minutes postprandial blood glucose, fasting blood glucose, high-density lipoprotein, age, body mass index, and glycated hemoglobin after inulin nutritional fortification intervention.
[0045] In one example, a machine learning model is built via XGBoost.
[0046] In one example, a machine learning model is established, and training the machine learning model according to the training variables includes:
[0047] Convert the training variables into DMatrix format, set the learning rate, maximum depth, and number of iterations, and obtain the trained model.
[0048] In one example, inputting the patient's training variables into the trained model to predict the patient's effect on inulin nutritional fortification intervention includes:
[0049] The SHAP threshold was set, the patient's training variables were input into the trained model, and the SHAP value of the patient was calculated. If the SHAP value was less than the SHAP threshold, the inulin nutritional fortification intervention was ineffective for the patient. If the SHAP value was greater than the SHAP threshold, the inulin nutritional fortification intervention was effective for the patient.
[0050] Figure 2a , Figure 2b Schematic diagrams of LASSO regression analysis of a LASSO coefficient path diagram and a cross curve are respectively shown according to an embodiment of the present invention.
[0051] Specifically, the initial variables were determined, which were the relevant information data of the included T2DM patients, including the patient's gender, age, body mass index (BMI), history of type 2 diabetes, fasting blood glucose (FBG), blood glucose values from 30 minutes to 180 minutes after a meal (30min, 60min, 120min, 180min), area under the curve of blood glucose changes (AUC), glycosylated hemoglobin (HbA1c), total cholesterol (TC), triglycerides (TG), low-density lipoprotein (LDL), high-density lipoprotein (HDL) and other information to train the model. Patients diagnosed with type 2 diabetes were given inulin nutritional fortification intervention for 12 weeks, twice a day, 10g / time, on the basis of the original drug intervention. The patients' glycosylated hemoglobin (HbA1c) levels were evaluated after the intervention, and patients with HbA1c <6.5% after the intervention were considered as the population with effective clinical diabetes remission achieved by inulin intervention. The initial variables were analyzed separately using the “CBCgrps” package in R, and the associations between these variables were analyzed using the “corrplot” package. The “glmnet” package in R was then used to perform the following analysis: Figure 2a , Figure 2b The LASSO regression (nfold=20) shown in the figure screens the training variables and changes the coefficients of the relevant independent variables to 0, thereby reducing the impact of multicollinearity on the regression results. The screening found that six indicators including glycated hemoglobin (HbA1c), the difference between fasting and 120 minutes after a meal (Delta PG), fasting blood glucose, high-density lipoprotein (HDL), age, and body mass index (BMI) can effectively predict the effect of inulin intervention, and the glycated hemoglobin after inulin nutritional fortification intervention constitutes the training variable.
[0052] XGBoost was used to build a machine learning model. The training variables were first converted into DMatrix format. Then the train() function was used from the caret package to optimize the parameters and output the best parameter configuration. Finally, the learning rate eta was set to 0.1, the maximum depth (max_depth) was set to 3, and the number of iterations was set to 100 to build the XGBoost model. The model-related performance was calculated using the "caret" package of R: accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score. Finally, the ROC curve, calibration curve, and clinical decision curve were plotted to analyze and evaluate the model. The model was explained by the SHAP value. Through this algorithm, after inputting the six indicators of T2DM patients, namely, glycated hemoglobin, blood sugar difference, fasting blood sugar, high-density lipoprotein, age, and body mass index, the SHAP value can be calculated, and the obtained value can be compared with the effective mean of 0.333. For SHAP values lower than 0.333, it is predicted that inulin intervention for 3 months is not enough to achieve clinical remission of diabetes with HbA1c < 6.5%, and for those higher than 0.333, inulin intervention may reduce HbA1c to the normal range below 6.5%. Therefore, through this algorithm, diabetic patients suitable for inulin intervention can be predicted and screened before intervention.
[0053] The present invention also provides an electronic device, which includes: a memory storing executable instructions; and a processor, which runs the executable instructions in the memory to implement the above-mentioned inulin nutrition fortification intervention effect prediction method.
[0054] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned inulin nutrition fortification intervention effect prediction method is implemented.
[0055] To facilitate understanding of the solutions and effects of the embodiments of the present invention, three specific application examples are given below. Those skilled in the art should understand that the examples are only for facilitating understanding of the present invention, and any specific details thereof are not intended to limit the present invention in any way.
[0056] Example 1
[0057] Figure 3a , Figure 3b , Figure 3c , Figure 3d , Figure 3e , Figure 3f Schematic diagrams of a training set ROC curve, a test set ROC curve, a training set calibration curve, a test set calibration curve, a training set decision curve, and a test set decision curve according to an embodiment of the present invention are respectively shown.
[0058] Figure 4A schematic diagram showing the importance of training variables using SHAP values according to an embodiment of the present invention is shown.
[0059] Figure 5 A schematic diagram of a SHAP variable dependency graph according to an embodiment of the present invention is shown.
[0060] The XGBoost prediction model showed that the accuracy of the training set was 0.819, the specificity was 0.913, the positive predictive value was 0.818, and the negative predictive value was 0.820; the accuracy of the test set was 0.709, the specificity was 0.909, and the positive predictive value was 0.710. The performance of the ROC curve analysis, calibration curve analysis, and decision curve analysis models is shown in Figure 2. Figure 3a-3f .like Figure 4 As shown in the figure, the SHAP values used to explain the model results show that the importance of the following six indicators in the training variables is ranked from high to low: glycated hemoglobin, the difference between fasting and 120 minutes after a meal, fasting blood glucose, high-density lipoprotein, age, body mass index, and the degree of dependence of these six training variables on the model, as shown in the figure. Figure 5 shown.
[0061] Figure 6 A schematic diagram of SHAP force graphs of two random patients according to an embodiment of the present invention is shown.
[0062] SHAP values allow for random inspection of the predictive value of any single sample in the sample and the impact of its features on the predictive value. Two patients are randomly selected, such as Figure 6 A and B in the figure use SHAP force diagram to explain their contribution to the feature factors in the model. From left to right, the value of each feature factor is shown in turn. The red bar graph represents the positive potential value of the patient, and the blue bar graph represents the negative potential value. E[f(x)] represents the output mean of the XGBoost model, and f(x) represents the SHAP value of a single sample. The SHAP force diagram of the first patient is shown in the figure below. Figure 6 As shown in Figure A, the effective mean of the XGBoost model for inulin intervention treatment is 0.333, while the SHAP value of the first patient reaches -0.150, which is lower than the effective value, indicating that inulin intervention treatment is ineffective for this patient. Similarly, the SHAP value of the second patient is 0.668, as shown in Figure Figure 6 As shown in Figure B, this is higher than the predicted mean of 0.333, indicating that the patient's type 2 diabetes condition was improved by inulin intervention treatment.
[0063] Example 2
[0064] The present disclosure provides an electronic device, which includes: a memory storing executable instructions; and a processor, which runs the executable instructions in the memory to implement the above-mentioned inulin nutrition fortification intervention effect prediction method.
[0065] An electronic device according to an embodiment of the present disclosure includes a memory and a processor.
[0066] The memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.
[0067] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory.
[0068] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.
[0069] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.
[0070] Example 3
[0071] An embodiment of the present disclosure provides a computer-readable storage medium storing a computer program, which implements the method for predicting the effect of inulin nutritional fortification intervention when executed by a processor.
[0072] According to the computer-readable storage medium of the embodiment of the present disclosure, non-transitory computer-readable instructions are stored thereon. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the above-mentioned methods of each embodiment of the present disclosure are executed.
[0073] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (e.g., memory card) and media with built-in ROM (e.g., ROM box).
[0074] Those skilled in the art should understand that the purpose of the above description of the embodiments of the present invention is only to exemplarily illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any given examples.
[0075] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.
Claims
1. A method for predicting the effect of inulin nutritional fortification intervention, characterized in that: include: Based on the patient information, determine the initial variables; Analyze the initial variables and screen the training variables; Establishing a machine learning model, training the machine learning model according to the training variables, and calculating a SHAP value; The patient's training variables were input into the trained model to predict the patient's effect on inulin nutritional fortification intervention.
2. The method for predicting the effect of inulin nutritional fortification intervention according to claim 1, wherein: The initial variables include the patient's gender, age, body mass index, history of type 2 diabetes, fasting blood glucose, blood glucose levels 30 minutes, 60 minutes, 120 minutes, and 180 minutes after a meal, the area under the curve of the postprandial blood glucose change curve, glycosylated hemoglobin, total cholesterol, triglycerides, low-density lipoprotein, high-density lipoprotein, and glycosylated hemoglobin after inulin nutritional fortification intervention.
3. The method for predicting the effect of inulin nutritional fortification intervention according to claim 1, wherein: The analysis of the initial variables includes: The initial variables were analyzed individually using the "CBCgrps" package, and the associations between these variables were analyzed using the "corrplot" package; The initial variables were selected according to LASSO regression in the "glmnet" package.
4. The method for predicting the effect of inulin nutritional fortification intervention according to claim 1, wherein: The training variables include glycated hemoglobin, the difference between fasting and 120 minutes postprandial blood glucose, fasting blood glucose, high-density lipoprotein, age, body mass index, and glycated hemoglobin after inulin nutritional fortification intervention.
5. The method for predicting the effect of inulin nutritional fortification intervention according to claim 1, wherein: Build machine learning models with XGBoost.
6. The method for predicting the effect of inulin nutritional fortification intervention according to claim 1, wherein: Establishing a machine learning model, wherein training the machine learning model according to the training variables comprises: Convert the training variables into DMatrix format, set the learning rate, maximum depth, and number of iterations, and obtain the trained model.
7. The method for predicting the effect of inulin nutritional fortification intervention according to claim 1, wherein: The patient's training variables were input into the trained model to predict the patient's effect on inulin nutritional fortification intervention, including: The SHAP threshold was set, the patient's training variables were input into the trained model, and the SHAP value of the patient was calculated. If the SHAP value was less than the SHAP threshold, the inulin nutritional fortification intervention was ineffective for the patient. If the SHAP value was greater than the SHAP threshold, the inulin nutritional fortification intervention was effective for the patient.
8. An electronic device, characterized in that: The electronic device comprises: A memory storing executable instructions; A processor, wherein the processor runs the executable instructions in the memory to implement the method for predicting the inulin nutritional fortification intervention effect according to any one of claims 1 to 7.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting the inulin nutritional fortification intervention effect according to any one of claims 1 to 7 is implemented.