Hypertensive nephropathy prediction method based on multi-model fusion screening characteristics

Abstract

The invention provides a hypertensive nephropathy prediction method based on multi-model fusion screening characteristics. An important feature set influencing the occurrence of hypertensive nephropathy is screened based on a multi-model fusion method, and a hypertensive nephropathy prediction model based on multi-model fusion is established by using the feature set. Early screening work for hypertension people is realized, so that the risk of hypertensive nephropathy can be found as early as possible, timely treatment is realized, and the prognosis of patients is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of hypertensive nephropathy prediction, and in particular to a hypertensive nephropathy prediction method based on multi-model fusion screening features. Background Art

[0002] Hypertension is a syndrome characterized by increased systemic arterial blood pressure, and its occurrence is related to lifestyle, eating habits and living environment. Long-term hypertension can cause arterial lesions and damage target organs, among which hypertensive nephropathy is more common. The prevalence of hypertensive nephropathy is increasing year by year. Therefore, hypertensive nephropathy is one of the main causes of ESRD. Hypertensive nephropathy may be asymptomatic in the early stages of onset, and may present with microalbuminuria, increased blood creatinine, etc. during the course of the disease, which can increase the risk of end-stage renal disease, adverse cardiovascular events and sudden death, posing a serious threat to the patient's life safety. Therefore, early screening, diagnosis and treatment of hypertensive patients are crucial.

[0003] In the prior art, on the one hand, when using clinically relevant information to predict the occurrence of hypertensive nephropathy, medical features are directly used to predict hypertensive nephropathy, or medical statistical methods and single machine learning algorithms are used to select risk factors that affect the occurrence of hypertensive nephropathy. However, the influencing factors are multidimensional, and some influencing factors have little effect on the occurrence of hypertensive nephropathy. Therefore, when predicting based on all medical features, the problem of large amount of calculation will be generated, and there will also be the problem of difficulty in data collection in screening applications. And using a single method to select important features is one-sided. On the other hand, for the prediction of hypertensive nephropathy in hypertensive people, most of the current studies use traditional statistical methods and traditional machine learning such as support vector machine (SVM), random forest (RF), etc., and the single prediction model has problems such as low prediction accuracy and weak generalization ability. Therefore, it is necessary to improve and merge the existing prediction models to improve the prediction accuracy and generalization ability of hypertensive nephropathy. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method for predicting hypertensive nephropathy based on multi-model fusion screening features.

[0005] To achieve the above object, the present invention adopts the following technical solution: a method for predicting hypertensive nephropathy based on multi-model fusion screening features, wherein the specific steps of the prediction method are as follows: S1: Collect relevant information of the hypertensive population, including hypertensive nephropathy patients and non-hypertensive nephropathy patients, and pre-process the relevant information to obtain initial data; The relevant information covers demographic characteristics, physical measurement characteristics, test characteristics, symptom characteristics, family history and medical history; Data preprocessing includes data anomalies, missing data, inconsistent data, incomplete data, and irregular data; S2: Obtain K sample sets through the K-fold cross-validation method, and divide them into training data sets and test data sets respectively; The initial data obtained in step S1 is divided into K sample subsets of equal size, and the K subsets are traversed in turn, each time the current subset is used as the test set and all the remaining samples are used as the training set to train and evaluate the model; the above operation forms K sample sets A1, A2...A k ; S3: Use multi-model fusion method to obtain important feature sets; On the K sample sets obtained in step S2, five machine learning multi-model fusion methods, including decision tree, random forest, XGBoost, AdaBoost, and GBDT, are used to obtain important feature sets; S4: Select the important feature set data consisting of N important features in step S3, and use the decision tree algorithm, random forest algorithm, XGBoost algorithm, AdaBoost algorithm, and GBDT algorithm respectively, and use the training set to train the five algorithms respectively; Obtain trained decision tree prediction model, random forest prediction model, XGBoost prediction model, AdaBoost prediction model, and GBDT prediction model; S5: using the test data set as the input of the five trained models, obtaining the prediction output results of the five models and performing weighted processing; integrating the weighted prediction output results in the K sample sets and dividing them into a training data set 1 and a test data set 1; S6: constructing a BP neural network, and training the BP neural network with the training data set 1 in step S5 to obtain a trained BP neural network; S7: using the test data set 1 as the input of the trained BP neural network to obtain a prediction output result 1 of the BP neural network; S8: Compare the prediction output result 1 with the true value in the test data set 1, and determine whether the fusion model composed of the decision tree model, random forest model, XGBoost model, AdaBoost model, GBDT model and BP neural network meets the prediction accuracy requirements according to the comparison result: If the prediction accuracy requirement is not met, repeat steps S4-S8 or repeat steps S6-S8; If the prediction accuracy requirement is met, the fusion model is used to predict the hypertension population.

[0006] Preferably, in step S3, five machine learning multi-model fusion methods are used to obtain the important feature set: S31: For each sample set A1, A2...A k , five machine learning algorithms, including decision tree, random forest, XGBoost, AdaBoost, and GBDT, were used to obtain the relative importance score and mean absolute SHAP value of each feature; The importance scores and mean absolute SHAP values ​​of the m features calculated using the above five machine learning algorithms form the matrix A; S32: According to the AUC values ​​of the five prediction models, the above matrix A is weighted to obtain a weighted relative importance score and a weighted average absolute SHAP value to form a matrix A1; S33: normalizing the calculated weighted relative importance score and weighted average absolute SHAP value to obtain a normalized index value B; S34: Calculate the average based on the normalized index value B to obtain the average index C of each feature; S35: Based on each sample set A1, A2...A in the K data sets k , obtain the total mean index of each feature; that is, K similar C matrix results are summarized into D matrix form; and the average of each row of D matrix is ​​calculated, that is, the total mean index E of each feature is obtained; S36: In the above-mentioned total mean index E of each feature, according to the preset number N of important features, obtain an important feature set.

[0007] Preferably, in step S1, data preprocessing includes the following specific operations: S11: Identify and process outliers in the data, and use statistical methods (such as Z-score or IQR) to eliminate or correct outliers; S12: Process missing data and fill missing values ​​using interpolation, mean filling or KNN filling; S13: Standardize inconsistent and irregular data, including coding of categorical variables and standardization of numerical variables; S14: Convert data from different data sources into a unified format to ensure data consistency.

[0008] Preferably, in step S8, the prediction accuracy includes multiple indicators, accuracy, recall rate, precision, F1-score and area under the ROC curve.

[0009] Preferably, in step S6, the structural optimization of the BP neural network includes using Dropout and BatchNormalization techniques to enhance the generalization capability of the model.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. The prediction method of the present invention obtains an important feature set based on the fusion method of five machine learning models, re-applies the K-fold cross-validation method, and uses a multi-model fusion learning method based on five machine learning algorithms to establish a hypertensive nephropathy screening model. This feature screening method based on multi-model fusion finds important features to establish a model, which is more conducive to improving the rationality and interpretability of the model.

[0011] 2. Use K data sets generated by the K-fold cross-validation method to traverse more data combinations, so that the important features screened out are universal and representative, and the generalization ability of the screening model is improved.

[0012] 3. Different models have different adaptability to data. By fusing multiple models, we can better cope with the diversity of data, improve the robustness of the model in different situations, integrate the advantages of different models, and improve performance and accuracy.

[0013] 4. The method proposed in this application can quickly obtain important features. It reduces the time and manpower cost required for modeling, and is conducive to improving the prediction efficiency and prediction accuracy of the model. In screening applications, it can reduce the cost of collecting data and improve the efficiency of disease screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a flow chart of a method for predicting hypertensive nephropathy based on multi-model fusion screening features; Figure 2 Flowchart for obtaining important feature sets for multi-model fusion methods. DETAILED DESCRIPTION

[0015] In order to provide a further understanding of the purpose, structure, features, and functions of the present invention, the following detailed description is given in conjunction with the embodiments.

[0016] Please refer to Figure 1 as well as Figure 2 The present invention provides a method for predicting hypertensive nephropathy based on multi-model fusion screening features, and the specific steps of the prediction method are as follows: S1: Collect relevant information of the hypertensive population, including hypertensive nephropathy patients and non-hypertensive nephropathy patients, and pre-process the relevant information to obtain initial data; The relevant information covers demographic characteristics, physical measurement characteristics, test characteristics, symptom characteristics, family history and medical history; Data preprocessing includes data anomalies, missing data, inconsistent data, incomplete data, and irregular data; S11: Identify and process outliers in the data, and use statistical methods (such as Z-score or IQR) to eliminate or correct outliers; S12: Process missing data and fill missing values ​​using interpolation, mean filling or KNN filling; S13: Standardize inconsistent and irregular data, including coding of categorical variables and standardization of numerical variables; S14: Convert data from different data sources into a unified format to ensure data consistency.

[0017] S2: Obtain K sample sets through the K-fold cross-validation method, and divide them into training data sets and test data sets respectively; The initial data obtained in step S1 is divided into K sample subsets of equal size, and the K subsets are traversed in turn, each time the current subset is used as the test set and all the remaining samples are used as the training set to train and evaluate the model; the above operation forms K sample sets A1, A2...A k ; S3: Use multi-model fusion method to obtain important feature sets; On the K sample sets obtained in step S2, five machine learning multi-model fusion methods, including decision tree, random forest, XGBoost, AdaBoost, and GBDT, are used to obtain important feature sets; S31: For each sample set A1, A2...A k , five machine learning algorithms, including decision tree, random forest, XGBoost, AdaBoost, and GBDT, were used to obtain the relative importance score and mean absolute SHAP value of each feature; The relative importance score is calculated using the scikit-learn library in Python to train machine learning models (decision trees, random forests, XGBoost, AdaBoost, GBDT), where the feature_importances_ attribute provides a direct quantitative indicator (feature importance score) to measure the influence of each feature on the model's prediction results.

[0018] The relative average absolute SHAP value is calculated using TreeExplainer in the SHAP library to calculate the SHAP value of the test set samples, showing the contribution of each feature to the prediction results. The importance of each feature is obtained by taking the average of the absolute values ​​of the SHAP value samples; Table 1 Eigenvalues ​​in each algorithm

[0019] Among them, ai1 represents the relative importance score of the ith feature calculated using the decision tree algorithm, and there are m medical feature variables in total. The same applies to other variables.

[0020] The importance scores and mean absolute SHAP values ​​of the m features calculated using the above five machine learning algorithms form the matrix A;

[0021] aij represents the importance score or average absolute SHAP value of the i-th feature calculated using the above five machine learning algorithms, where i is a natural number from 1 to m, and j is a natural number from 1 to 10.

[0022] S32: According to the AUC values ​​of the five prediction models, the above matrix A is weighted to obtain a weighted relative importance score and a weighted average absolute SHAP value to form a matrix A1;

[0023] Among them, w1 represents the AUC value of the decision tree model on the test set, w2 represents the AUC value of the random forest model on the test set, w3 represents the AUC value of the XGBoost model on the test set, w4 represents the AUC value of the AdaBoost model on the test set, and w5 represents the AUC value of the GBDT model on the test set.

[0024] S33: normalizing the calculated weighted relative importance score and weighted average absolute SHAP value to obtain a normalized index value B; Min-Max normalization is also called linear normalization or maximum and minimum normalization, which scales the data to the interval [0,1]. This is one of the simplest normalization methods; formula: ; Among them, X is the original data, Xmin and Xmax are the minimum and maximum values ​​in the data set respectively, and Xnorm is the normalized data.

[0025] The A1 matrix is ​​converted into the normalized index value B through the above formula; ; in , That is, the minimum value in the jth column of the A1 matrix, That is, the maximum value in the j-th column of the A1 matrix.

[0026] S34: Calculate the average based on the normalized index value B to obtain the average index C of each feature;

[0027] S35: Based on each sample set A1, A2...A in the K data sets k, obtain the total mean index of each feature; that is, K similar C matrix results are summarized into D matrix form;

[0028] Where ciq represents the mean index of the ith feature in the Aqth sample set. i is a natural number from 1 to m, and q is a natural number from 1 to K.

[0029] And find the mean of each row of the D matrix, that is, obtain the total mean index E of each feature;

[0030] Among them, ei represents the total mean index of feature i.

[0031] S36: In the above-mentioned total mean index E of each feature, according to the preset number N of important features, obtain an important feature set.

[0032] In an embodiment of the present invention, feature selection is the selection of the number of important features, and the total mean index value of each feature is sorted in order from large to small; the candidate features corresponding to the total mean index values ​​of the top N positions in the sort are intercepted to obtain a set of important features, where N is the number corresponding to the quantity selection strategy.

[0033] The choice of N is adjusted according to the accuracy of the final fusion model.

[0034] S4: In step S3, select important feature set data consisting of N important features, and use decision tree algorithm, random forest algorithm, XGBoost algorithm, AdaBoost algorithm, GBDT algorithm respectively, and use the training set to train the five algorithms respectively.

[0035] Obtain the trained decision tree prediction model, random forest prediction model, XGBoost prediction model, AdaBoost prediction model, and GBDT prediction model.

[0036] Using the sklearn package of the python development language, five algorithms, decision tree, random forest, XGBoost, AdaBoost, and GBDT, were selected to establish the hypertensive nephropathy screening model respectively; the screening parameter set that can distinguish hypertensive nephropathy was established, and the data was trained and iteratively trained in the five models, and the hypertensive nephropathy screening parameter set was tuned, and finally the hypertensive nephropathy screening parameter set that can distinguish hypertensive nephropathy was obtained. The optimal hyperparameter combination of each model during the training process was obtained by the grid search method.

[0037] S5: using the test data set as the input of the five trained models, obtaining the prediction output results of the five models and performing weighted processing; integrating the weighted prediction output results in the K sample sets and dividing them into a training data set 1 and a test data set 1; In this embodiment, the sample subsets A1, A2...A k In the sample subset, the prediction output results m1, m2, m3, m4, and m5 of the five models on the test data set (n rows of data) are integrated into n rows and 5 columns of data Mn*5, and then weighted calculation is performed according to the number of samples predicted correctly by each model. Similarly, the K sample sets, that is, K results, are integrated into K*n rows and 5 columns of data as the new feature set P, which is divided into training data set one and test data set one.

[0038] The output result m1: the prediction result of the decision tree prediction model on the test data set, where mij is the data in the i-th row, and the probability value of the true label 1 (hypertensive nephropathy) of the output sample is in the range of 0 to 1, and the values ​​of m2, m3, m4, and m5 are similar;

[0039] The Mn*5 matrix is:

[0040] The Nn*5 matrix is:

[0041] Where w1 is the number of samples correctly predicted by the decision tree model / the sum of the number of samples correctly predicted by the above five models*100%. w2, w3, w4, w5 and so on. For K sample sets, K N are generated, namely: N1, N2, N3...NK.

[0042] The P matrix is: .

[0043] S6: constructing a BP neural network, and training the BP neural network with the training data set 1 in step S5 to obtain a trained BP neural network; The BP algorithm is an error back propagation algorithm. It is mainly divided into two parts: forward propagation and back propagation. In forward propagation, the output of each neuron is calculated layer by layer from the input layer to the output layer through the weighted connection between neurons to obtain the predicted value. In back propagation, the error of each layer is calculated based on the difference between the predicted value and the true value, and the parameters of the network are adjusted by the gradient descent method to minimize the error. This process is repeated many times until the error of the network is less than a certain threshold or other termination conditions are met.

[0044] The present invention adopts BP neural network to fuse the prediction outputs of decision tree model, random forest model, XGBoost model, AdaBoost model and GBDT model, and can use its excellent fitting ability to fuse model results and output more accurate prediction values. At the same time, its simple structure will not increase the complexity of the model, so that the fusion model has higher stability.

[0045] S7: using the test data set 1 as the input of the trained BP neural network to obtain a prediction output result 1 of the BP neural network; S8: Compare the prediction output result 1 with the true value in the test data set 1, and determine whether the fusion model composed of the decision tree model, random forest model, XGBoost model, AdaBoost model, GBDT model and BP neural network meets the prediction accuracy requirements according to the comparison result: If the prediction accuracy requirement is not met, repeat steps S4-S8 or repeat steps S6-S8; If the prediction accuracy requirement is met, the fusion model is used to predict the hypertension population.

[0046] In a specific embodiment, when comparing the predicted output result 1 with the true value in the test data set 1, the AUC value is used as an evaluation indicator.

[0047] In this embodiment, an AUC value prediction error threshold is set according to the prediction accuracy requirement. When the AUC value of the fusion model obtained through training is greater than or equal to the AUC value prediction error threshold, the fusion model is the final fusion model.

[0048] The AUC value mentioned above, AUC is the area under the ROC curve. The ROC curve uses the True Positive Rate (TPR) and False Positive Rate (FPR) of the hypertensive nephropathy screening model in the test data as the horizontal and vertical coordinates to depict the performance of the screening model at different thresholds; the AUC value is used to measure the performance of the screening model. The closer the AUC value is to 1, the better the performance of the screening model; conversely, the closer the AUC value is to 0, the worse the performance of the screening model. Through the ROC curve, we can intuitively understand the performance of the screening model at different thresholds; and through the AUC value, we can quantitatively evaluate the overall performance of the screening model.

[0049] Preferably, in step S8, the prediction accuracy includes multiple indicators, accuracy, recall rate, precision, F1-score and area under the ROC curve.

[0050] The accuracy rate is the ratio of the number of correctly predicted samples to the total number of samples. Its calculation formula is: Accuracy rate = number of correctly predicted samples / total number of samples; Recall rate, also known as true positive rate, indicates the proportion of positive samples that the model can correctly identify to all positive samples. Its calculation formula is: Recall rate = true positive samples / (true positive samples + false negative samples); Precision refers to the proportion of samples predicted by the model as positive samples that are actually positive samples. Its calculation formula is: Precision = number of true samples / (number of true samples + number of false positive samples); F1-score is the harmonic average of precision and recall, and is often used to balance the weights of the two. Its calculation formula is: F1=2×(precision×recall) / (precision+recall); The ROC curve (Receiver Operating Characteristic Curve) plots the relationship between the false positive rate (FPR) and the true positive rate (TPR). By changing the classification threshold, different FPRs and TPRs can be obtained, thus forming a curve.

[0051] AUC (Area Under the Curve) is the area under this curve and ranges between 0 and 1. AUC = 1: Perfect model. AUC = 0.5: Random guessing model. AUC < 0.5: Model performs worse than random guessing.

[0052] Preferably, in step S6, the structural optimization of the BP neural network includes using Dropout and BatchNormalization techniques to enhance the generalization capability of the model.

[0053] The present invention has been described by the above-mentioned relevant embodiments, however, the above-mentioned embodiments are only examples for implementing the present invention. It must be pointed out that the disclosed embodiments do not limit the scope of the present invention. On the contrary, changes and modifications made without departing from the spirit and scope of the present invention are all within the scope of patent protection of the present invention.

Claims

1. A method for predicting hypertensive nephropathy based on multi-model fusion screening features, characterized in that: The specific steps of the prediction method are as follows: S1: Collect relevant information of the hypertensive population, including hypertensive nephropathy patients and non-hypertensive nephropathy patients, and pre-process the relevant information to obtain initial data; The relevant information covers demographic characteristics, physical measurement characteristics, test characteristics, symptom characteristics, family history and medical history; Data preprocessing includes data anomalies, missing data, inconsistent data, incomplete data, and irregular data; S2: Obtain K sample sets through the K-fold cross-validation method, and divide them into training data sets and test data sets respectively; The initial data obtained in step S1 is divided into K sample subsets of equal size, and the K subsets are traversed in turn, each time the current subset is used as the test set and all the remaining samples are used as the training set to train and evaluate the model; the above operation forms K sample sets A1, A2...A k ; S3: Use multi-model fusion method to obtain important feature sets; On the K sample sets obtained in step S2, five machine learning multi-model fusion methods, including decision tree, random forest, XGBoost, AdaBoost, and GBDT, are used to obtain important feature sets; S4: Select the important feature set data consisting of N important features in step S3, and use the decision tree algorithm, random forest algorithm, XGBoost algorithm, AdaBoost algorithm, and GBDT algorithm respectively, and use the training set to train the five algorithms respectively; Obtain trained decision tree prediction model, random forest prediction model, XGBoost prediction model, AdaBoost prediction model, and GBDT prediction model; S5: using the test data set as the input of the five trained models, obtaining the prediction output results of the five models and performing weighted processing; integrating the weighted prediction output results in the K sample sets and dividing them into a training data set 1 and a test data set 1; S6: constructing a BP neural network, and training the BP neural network with the training data set 1 in step S5 to obtain a trained BP neural network; S7: using the test data set 1 as the input of the trained BP neural network to obtain a prediction output result 1 of the BP neural network; S8: Compare the prediction output result 1 with the true value in the test data set 1, and determine whether the fusion model composed of the decision tree model, random forest model, XGBoost model, AdaBoost model, GBDT model and BP neural network meets the prediction accuracy requirements according to the comparison result: If the prediction accuracy requirement is not met, repeat steps S4-S8 or repeat steps S6-S8; If the prediction accuracy requirement is met, the fusion model is used to predict the hypertension population.

2. A method for predicting hypertensive nephropathy based on multi-model fusion screening features as claimed in claim 1, characterized in that: In step S3, five machine learning multi-model fusion methods are used to obtain the important feature set as follows: S31: For each sample set A1, A2...A k , five machine learning algorithms, including decision tree, random forest, XGBoost, AdaBoost, and GBDT, were used to obtain the relative importance score and mean absolute SHAP value of each feature; The importance scores and mean absolute SHAP values ​​of the m features calculated using the above five machine learning algorithms form the matrix A; S32: According to the AUC values ​​of the five prediction models, the above matrix A is weighted to obtain a weighted relative importance score and a weighted average absolute SHAP value to form a matrix A1; S33: normalizing the calculated weighted relative importance score and weighted average absolute SHAP value to obtain a normalized index value B; S34: Calculate the average based on the normalized index value B to obtain the average index C of each feature; S35: Based on each sample set A1, A2...A in the K data sets k , obtain the total mean index of each feature; that is, K similar C matrix results are summarized into D matrix form; and the average of each row of D matrix is ​​calculated, that is, the total mean index E of each feature is obtained; S36: In the above-mentioned total mean index E of each feature, according to the preset number N of important features, obtain an important feature set.

3. The method for predicting hypertensive nephropathy based on multi-model fusion screening features as claimed in claim 1, characterized in that: In step S1, data preprocessing includes the following specific operations: S11: Identify and process outliers in the data, and use statistical methods to eliminate or correct outliers; S12: Process missing data and fill missing values ​​using interpolation, mean filling or KNN filling; S13: Standardize inconsistent and irregular data, including coding of categorical variables and standardization of numerical variables; S14: Convert data from different data sources into a unified format to ensure data consistency.

4. The method for predicting hypertensive nephropathy based on multi-model fusion screening features as claimed in claim 1, characterized in that: In step S8, the prediction accuracy includes multiple indicators, accuracy, recall, precision, F1-score and area under the ROC curve.

5. The method for predicting hypertensive nephropathy based on multi-model fusion screening features as claimed in claim 1, characterized in that: In step S6, the structural optimization of the BP neural network includes using Dropout and Batch Normalization techniques to improve the generalization ability of the model.

Images