Diabetic nephropathy classification model construction method and device based on fuzzy decision-making system

Abstract

The invention discloses a diabetic nephropathy classification model construction method and device based on a fuzzy decision system. In the process of constructing a diabetic nephropathy classification model, a feature group generation strategy based on a fuzzy approximate Markov blanket (FAMB) is determined specifically by introducing a Markov blanket algorithm and approximation and fuzzy uncertainty measurement of a fuzzy rough set; further dividing the overlapped features based on an affiliation division mechanism of a fuzzy dependency degree; and finally, determining a heterogeneous feature selection algorithm with group structure mining, and screening and optimizing feature groups. According to the method, a measurement index with high calculation efficiency can be developed in a centralized manner, and the classification performance is improved by further utilizing the structural information between the features, so that a reference is provided for clinical auxiliary diagnosis.

Description

Technical Field

[0001] The present invention relates to the field of neural network technology, and in particular to a method and device for constructing a diabetic nephropathy classification model based on a fuzzy decision system. Background Art

[0002] Given the large differences in treatment and prognosis between diabetic nephropathy (DN) and non-diabetic kidney disease (NDRD), the existing and publicly available solution is to diagnose and determine through "renal puncture biopsy". However, due to its severe invasiveness and postoperative complications, this technology is limited in clinical practice. Therefore, using appropriate technology to capture important pathological attributes and mining biomarkers containing multiple attributes is of great significance for evaluating DN and establishing diagnostic prediction models through non-invasive diagnostic methods. Summary of the invention

[0003] Based on this, the embodiment of the present application provides a method and device for constructing a diabetic nephropathy classification model based on a fuzzy decision system, which can focus on developing a metric with high computational efficiency, and further utilize the structural information between features to improve the classification performance, thereby providing a reference for clinical auxiliary diagnosis.

[0004] In a first aspect, a method for constructing a diabetic nephropathy classification model based on a fuzzy decision system is provided, the method comprising:

[0005] According to the target diabetic nephropathy data set, the diabetic nephropathy detection indicators were analyzed and determined; the data corresponding to each diabetic nephropathy detection indicator was imported into SPSS software, and the data was preprocessed; and the missing values ​​in the data set were processed using the adjacent point average method; according to the preprocessed biopsy data, the data was divided into a DN group data set and a NDRD group data set;

[0006] Constructing a diabetic nephropathy classification model; wherein the diabetic nephropathy classification model specifically includes introducing the Markovblanket algorithm and the approximation of fuzzy rough sets and fuzzy uncertainty measurement, determining a feature group generation strategy based on FAMB; further dividing overlapping features based on the attribution division mechanism of fuzzy dependency; and finally determining a heterogeneous feature selection algorithm with group structure mining to screen and optimize the feature group;

[0007] The diabetic nephropathy classification model is trained by using the divided DN group data set and the NDRD group data set until a trained diabetic nephropathy classification model is obtained.

[0008] Optionally, the method further comprises:

[0009] The constructed diabetic nephropathy classification model was compared with other feature selection algorithm models; among which, other feature selection algorithm models at least included DRGS, FastCFS, FCBF, IWFS, MRI, MRMI and N-MRMCR-MI algorithms;

[0010] And the classification results are determined by multiple classifiers; wherein the classifiers include at least Classification And Regression Tree, Bayes, K-Nearest Neighbor, and Support Vector Machine classifiers;

[0011] The evaluation results of each model were obtained based on classification accuracy and average reduction rate.

[0012] Optionally, the model comparison and evaluation is implemented through Matlab software, and different algorithm models are generated to obtain a relationship diagram between the number of selected features and classification accuracy.

[0013] Optionally, before training the diabetic nephropathy classification model using the divided DN group data set and NDRD group data set, the method further includes:

[0014] Split the dataset into training / testing datasets using cross-validation.

[0015] Optionally, according to the target diabetic nephropathy data set, the diabetic nephropathy detection indicators determined by analysis specifically include:

[0016] The patient's ID, physical parameters, diagnosis results, prealbumin, creatine kinase, urea, apolipoprotein A1, low-density lipoprotein cholesterol, albumin / globulin, glomerular filtration rate, alkaline phosphatase, apolipoprotein B, amylase, α-hydroxybutyrate dehydrogenase, V-glutamyl transpeptidase, indirect bilirubin, total calcium, potassium, and chloride.

[0017] In a second aspect, a device for constructing a diabetic nephropathy classification model based on a fuzzy decision system is provided, the device comprising:

[0018] The preprocessing module is used to analyze and determine the diabetic nephropathy detection indicators according to the target diabetic nephropathy data set; import the data corresponding to each diabetic nephropathy detection indicator into SPSS software and preprocess the data; and use the adjacent point average method to process the missing values ​​in the data set; according to the preprocessed biopsy data, the data is divided into a DN group data set and a NDRD group data set;

[0019] A construction module is used to construct a diabetic nephropathy classification model; wherein the diabetic nephropathy classification model specifically includes introducing a Markov blanket algorithm and fuzzy rough set approximation and fuzzy uncertainty measurement, determining a feature group generation strategy based on FAMB; further dividing overlapping features based on a fuzzy dependency attribution division mechanism; and finally determining a heterogeneous feature selection algorithm with group structure mining to screen and optimize feature groups;

[0020] The training module is used to train the diabetic nephropathy classification model through the divided DN group data set and the NDRD group data set until a trained diabetic nephropathy classification model is obtained.

[0021] Optionally, the device further comprises:

[0022] A comparison module is used to compare the constructed diabetic nephropathy classification model with other feature selection algorithm models; wherein the other feature selection algorithm models include at least DRGS, FastCFS, FCBF, IWFS, MRI, MRMI and N-MRMCR-MI algorithms; and to determine the classification results through multiple classifiers; wherein the classifiers include at least ClassificationAndRegressionTree, Bayes, K-Nearest Neighbor and SupportVectorMachine classifiers; the evaluation results of each model are obtained based on classification accuracy and average reduction rate.

[0023] Optionally, the model comparison and evaluation is implemented through Matlab software, and different algorithm models are generated to obtain a relationship diagram between the number of selected features and classification accuracy.

[0024] Optionally, before training the diabetic nephropathy classification model using the divided DN group data set and NDRD group data set, the method further includes:

[0025] Split the dataset into training / testing datasets using cross-validation.

[0026] Optionally, according to the target diabetic nephropathy data set, the diabetic nephropathy detection indicators determined by analysis specifically include:

[0027] The patient's ID, physical parameters, diagnosis results, prealbumin, creatine kinase, urea, apolipoprotein A1, low-density lipoprotein cholesterol, albumin / globulin, glomerular filtration rate, alkaline phosphatase, apolipoprotein B, amylase, α-hydroxybutyrate dehydrogenase, V-glutamyl transpeptidase, indirect bilirubin, total calcium, potassium, and chloride.

[0028] The technical solution provided in the embodiment of the present application constructs a classification model for diabetic nephropathy, specifically by introducing the Markovblanket algorithm and the approximation of fuzzy rough sets and fuzzy uncertainty measurement, a feature group generation strategy based on FAMB is determined; overlapping features are further divided based on the attribution division mechanism of fuzzy dependency; finally, a heterogeneous feature selection algorithm with group structure mining is determined to screen and optimize the feature group. It can be seen that the beneficial effect of the present invention is that it can focus on developing a metric with high computational efficiency, further utilize the structural information between features to improve the classification performance, and thus provide a reference for clinical auxiliary diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0030] Figure 1 A flowchart for constructing a diabetic nephropathy classification model based on a fuzzy decision system provided in an embodiment of the present application;

[0031] Figure 2 A flowchart of the HFS-GSM algorithm provided in an embodiment of the present application;

[0032] Figure 3 This is a framework diagram of the HFS-GSM algorithm provided in the embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0034] In the description of the present invention, the terms "comprises", "has" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may also include other steps or units that are not explicitly listed but are inherent to these processes, methods, products or apparatuses, or steps or units added based on further optimization schemes conceived by the present invention.

[0035] Given the large differences in treatment and prognosis between diabetic nephropathy (DN) and non-diabetic kidney disease (NDRD), the existing and publicly available solution is to diagnose and determine through "renal puncture biopsy". However, due to its severe invasiveness and postoperative complications, this technology is limited in clinical practice. Therefore, using appropriate technology to capture important pathological attributes and mining biomarkers containing multiple attributes is of great significance for evaluating DN and establishing diagnostic prediction models through non-invasive diagnostic methods.

[0036] Defects in the extraction of existing diabetic nephropathy detection indicators: By using different classifiers in 12 public data sets and 7 other representative feature selection algorithm models to extract diabetic nephropathy detection indicators, it was found that the defect of the existing solution was that MRMI obtained the best simplification results when tested with four different classifiers. However, this was at the expense of a significant sacrifice in classification accuracy. Similarly, the FCBF algorithm with output subsets did not achieve a good compromise between classification accuracy and the number of selected features, although its classification accuracy was slightly better than the MRMI algorithm in some cases. The remaining other algorithms have the problem of large original space redundancy.

[0037] Specifically, please refer to Figure 1 , which shows a flow chart of a method for constructing a diabetic nephropathy classification model based on a fuzzy decision system provided in an embodiment of the present application. The method may include the following steps:

[0038] Step 101: Analyze and determine diabetic nephropathy detection indicators based on the target diabetic nephropathy data set.

[0039] The data corresponding to each diabetic nephropathy detection index were imported into SPSS software, and the data were preprocessed; the missing values ​​in the data set were processed using the adjacent point average method; based on the preprocessed biopsy data, the data were divided into DN group data set and NDRD group data set.

[0040] Step 102, constructing a diabetic nephropathy classification model.

[0041] Among them, the diabetic nephropathy classification model specifically includes the introduction of Markov blanket algorithm and fuzzy rough set approximation and fuzzy uncertainty measurement to determine the feature group generation strategy based on FAMB; further divide the overlapping features based on the attribution division mechanism of fuzzy dependency; finally determine the heterogeneous feature selection algorithm with group structure mining to screen and optimize the feature group.

[0042] Step 103: training the diabetic nephropathy classification model using the divided DN group data set and the NDRD group data set until a trained diabetic nephropathy classification model is obtained.

[0043] The heterogeneous feature selection method based on fuzzy decision system (HFS-GSM) in this application follows the principle of a "strategy" and a "mechanism". (1) In the strategy: the concept of Markov blanket is introduced in fuzzy rough set, and a feature group generation strategy based on fuzzy approximate Markov blanket (FAMB) is designed by using the ideas of approximation and fuzzy uncertainty measurement. (2) Mechanism: In view of the overlapping of feature groups, a classification mechanism based on fuzzy dependency is proposed.

[0044] The HFS-GSM framework is divided into three parts: first, the operation mechanism of the feature group generator is based on the concept of fuzzy approximate Markov blanket with fuzzy uncertainty measure, and a feature group generation strategy with group structure mining is planned; second, for the case of overlapping feature groups, an overlapping group elimination mechanism is developed, in which fuzzy dependencies are used to partition the features in the overlapping area to determine which group they belong to; finally, in the feature group selection stage, a heterogeneous feature selection algorithm with group structure mining is designed.

[0045] Using the ideas of approximation and Markov blanket, this method is not only able to mine information using group structure, but also improves computational efficiency to a certain extent. This is due to the fact that it replaces the returns from global calculations based on probability calculations. In addition, a mechanism for eliminating overlapping groups is established using fuzzy dependencies to avoid possible redundancy in the computational process. The superiority and effectiveness of this method are demonstrated by comparing it with 7 representative feature selection methods in 12 publicly available medical datasets. More importantly, the effectiveness and feasibility of this method in medical diagnosis are further illustrated from practical applications by examining a real medical case of diabetic nephropathy. It provides a good solution for doctors to conduct clinical auxiliary diagnosis.

[0046] The method in this application also includes a comparison process with other algorithms. A specific embodiment is given below:

[0047] (1) Experimental data preprocessing

[0048] ①Preliminary preparation

[0049] Dataset: Based on the diabetic nephropathy data set provided by the hospital, diabetic nephropathy detection indicators were analyzed, such as: patient ID, physical indicators, diagnosis results, prealbumin, creatine kinase, urea, urea, apolipoprotein A1, low-density lipoprotein cholesterol, low-density lipoprotein cholesterol, albumin / globulin, glomerular filtration rate (eGFR), alkaline phosphatase (AKP), apolipoprotein B (Apo), amylase (AMY), α-hydroxybutyrate dehydrogenase, V-glutamyl transpeptidase, indirect bilirubin, total calcium, potassium, chloride, etc.;

[0050] ② Experimental process

[0051] Import the data set into SPSS software; first pre-process the data with many duplicates and missing values; then use the "average value of adjacent points" method to process the missing values ​​in the data set.

[0052] ③ Experimental results

[0053] Complete medical records of 1367 patients were obtained, including 54 test indicators. According to the pathological biopsy results, the patients were divided into DN group (127 cases) and NDRD group (1240 cases).

[0054] (2) Screening characteristic indicators

[0055] ①Preparation of experimental data and relevant specifications

[0056] Dataset: valid data set after experimental preprocessing;

[0057] Algorithm models: DRGS, FastCFS, FCBF, IWFS, MRI, MRMI, N-MRMCR-MI, HFS-GSM and other feature selection algorithm models;

[0058] Evaluation criteria: Classification accuracy is based on four classifiers: Classification And Regression Tree (CART), Bayes (NB), K-Nearest Neighbor (KNN), Support Vector Machine (SVM), the classification accuracy and average reduction rate of the eight algorithms on the DN dataset when tested on four different classifiers are set as evaluation indicators. Experimental regulations: the higher the classification accuracy and average reduction rate, the better the algorithm effect;

[0059] ② Experimental process

[0060] Select feature indicators: put the data sets into the feature selection algorithm for feature selection;

[0061] Algorithm effect evaluation: Evaluate each feature selection algorithm according to the agreed algorithm effect evaluation specifications;

[0062] Select the optimal algorithm: Based on the algorithm effect evaluation, the optimal algorithm "HFS-GSM algorithm" is selected, with a classification accuracy of "91.95+0.24" and an average reduction rate of "74.0175%";

[0063] ③ Experimental results

[0064] The experimental results show that the following indicators "classification accuracy, average simplification rate" are selected as characteristic indicators.

[0065] The technical principle of this embodiment is as follows: the concept of Markov blanket is introduced in fuzzy rough set, and a feature group generation strategy based on fuzzy approximate Markov blanket (FAMB) is designed by using the idea of ​​approximation and fuzzy uncertainty measurement. In addition, a fuzzy dependency-based attribution division mechanism is proposed to solve the problem of feature group overlap.

[0066] By using different classifiers in 12 public medical data sets and 7 other representative feature selection algorithm models to extract diabetic nephropathy detection indicators, we then built an algorithm model for predicting early-stage diabetic nephropathy patients and compared the predictive performance of each algorithm model. We then selected the model with the best predictive performance to assist clinical decision-making.

[0067] In the process of implementing this study, a large number of experiments were conducted to prove the data dimensionality reduction performance based on DRGS, FastCFS, FCBF, IWFS, MRI, MRMI, N-MRMCR-MI and HFS-GSM algorithm models, and to evaluate the accuracy of diabetic nephropathy feature measurement indicators extracted by several algorithm models. The experimental data are divided into 12 public medical data sets (audiology, cardiotolography, cleveland, covid-19, dermatology, HCV, hepatitis, lung cancer, lymphography, parkinsons, SPECT, thyroid disease) and diabetic nephropathy patient classification (DN) data sets. These data have a wide range of practical applications and meet multiple conditions such as different sources, different categories, different features, and different sample numbers. The data set includes discrete features and continuous features. In the experiment, fuzzy rough set tools are used to simultaneously reduce the data dimensionality of discrete and continuous features.

[0068] Experimental settings: This experiment uses 8 feature selection algorithms for comparison, namely DRGS, FastCFS, FCBF, IWFS, MRI, MRMI, N-MRMCR-MI and HFS-GSM algorithms. According to the different characteristics of the algorithms and experimental requirements, the corresponding parameters are set. For the setting of the classifier, 4 different classifiers are selected here (ClassificationAndRegressionTree(CART), Bayes (NB), K-Nearest Neighbor (KNN), Support Vector Machine (SVM)) First, the data set is divided into training / testing data sets through cross validation. Secondly, the feature subsets are judged and set. Next, the classifier is trained on the training data set, and then the test data is input into the classifier to obtain the classification accuracy. Finally, different algorithm models are used to obtain the relationship between the number of selected features and the classification accuracy. The experiment is carried out based on Matlab software.

[0069] The experimental results were evaluated by classification accuracy and average reduction rate. Four different classifiers were used to evaluate the classification accuracy. The professional measurement indicators given by medical experts and the average reduction rate extracted based on the algorithm model represent the matching rate of the feature selection results.

[0070] Model feature extraction was performed on the benchmark dataset and diabetic nephropathy patient classification dataset. The experimental results are as follows:

[0071] The classification accuracy tables of various algorithms based on four classifiers (CART, NB, KNN, SVM) (Table 1, Table 2, Table 3, Table 4). Since the feature selection algorithm ranks the weights of each indicator (Table 5), by calculating the average simplification rate (%) of the eight algorithms on all data sets and the DN data set when tested using four different classifiers, the average simplification rates calculated by the HFS-GSM algorithm reached 70%, 69%, 65%, 63% and 73.1%, 79.52%, 73.1%, 70.35% respectively. It can be seen that although the classification accuracy of the features extracted under the HFS-GSM algorithm is not as good as that of the MRMI algorithm, it also significantly reduces the consumption of computing resources and ensures the overall classification performance. That is, the feature indicators are: classification accuracy "91.95+0.24", average simplification rate "74.0175%".

[0072] Table 1 The classification performance of different algorithms with CART

[0073]

[0074] Table 2 The classification performance of different algorithms with NB

[0075]

[0076] Table 3 The classification performance of different algorithms with KNN

[0077]

[0078] Table 4 The classification performance of different algorithms with SVM

[0079]

[0080] Table 5 The classification performance of different algorithms on DN dataset

[0081]

[0082] The present application also provides a device for constructing a diabetic nephropathy classification model based on a fuzzy decision system, comprising:

[0083] The preprocessing module is used to analyze and determine the diabetic nephropathy detection indicators according to the target diabetic nephropathy data set; import the data corresponding to each diabetic nephropathy detection indicator into SPSS software and preprocess the data; and use the adjacent point average method to process the missing values ​​in the data set; according to the preprocessed biopsy data, the data is divided into a DN group data set and a NDRD group data set;

[0084] A construction module is used to construct a diabetic nephropathy classification model; the diabetic nephropathy classification model specifically includes the introduction of the Markov blanket algorithm and the approximation of fuzzy rough sets and fuzzy uncertainty measurement, and the determination of a feature group generation strategy based on FAMB; further, the overlapping features are divided based on the attribution division mechanism of fuzzy dependency; finally, a heterogeneous feature selection algorithm with group structure mining is determined to screen and optimize the feature group;

[0085] The training module is used to train the diabetic nephropathy classification model through the divided DN group data set and the NDRD group data set until a trained diabetic nephropathy classification model is obtained.

[0086] In an optional embodiment of the present application, the device further includes:

[0087] A comparison module is used to compare the constructed diabetic nephropathy classification model with other feature selection algorithm models; wherein the other feature selection algorithm models include at least DRGS, FastCFS, FCBF, IWFS, MRI, MRMI and N-MRMCR-MI algorithms; and to determine the classification results through multiple classifiers; wherein the classifiers include at least ClassificationAndRegressionTree, Bayes, K-Nearest Neighbor and SupportVectorMachine classifiers; the evaluation results of each model are obtained based on classification accuracy and average reduction rate.

[0088] In an optional embodiment of the present application, model comparison and evaluation is implemented through Matlab software, and different algorithm models are generated to obtain a relationship diagram between the number of selected features and classification accuracy.

[0089] In an optional embodiment of the present application, before training the diabetic nephropathy classification model using the divided DN group data set and NDRD group data set, the method further includes:

[0090] Split the dataset into training / testing datasets using cross-validation.

[0091] In an optional embodiment of the present application, according to the target diabetic nephropathy data set, the diabetic nephropathy detection index determined by analysis specifically includes:

[0092] The patient's ID, physical parameters, diagnosis results, prealbumin, creatine kinase, urea, apolipoprotein A1, low-density lipoprotein cholesterol, albumin / globulin, glomerular filtration rate, alkaline phosphatase, apolipoprotein B, amylase, α-hydroxybutyrate dehydrogenase, V-glutamyl transpeptidase, indirect bilirubin, total calcium, potassium, and chloride.

[0093] The device for constructing a diabetic nephropathy classification model based on a fuzzy decision system provided in the embodiment of the present application is used to implement the above-mentioned method for constructing a diabetic nephropathy classification model based on a fuzzy decision system. For the specific definition of the device for constructing a diabetic nephropathy classification model based on a fuzzy decision system, please refer to the definition of the method for constructing a diabetic nephropathy classification model based on a fuzzy decision system in the above text, which will not be repeated here. The various parts of the above-mentioned device for constructing a diabetic nephropathy classification model based on a fuzzy decision system can be fully or partially implemented by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the device in the form of hardware, or can be stored in the memory in the device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0094] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A method for constructing a diabetic nephropathy classification model based on a fuzzy decision system, characterized in that: The method comprises: According to the target diabetic nephropathy data set, the diabetic nephropathy detection indicators were analyzed and determined; the data corresponding to each diabetic nephropathy detection indicator was imported into SPSS software, and the data was preprocessed; and the missing values ​​in the data set were processed using the adjacent point average method; according to the preprocessed biopsy data, the data was divided into a DN group data set and a NDRD group data set; Constructing a diabetic nephropathy classification model; wherein the diabetic nephropathy classification model specifically includes introducing the Markovblanket algorithm and the approximation of fuzzy rough sets and fuzzy uncertainty measurement, determining a feature group generation strategy based on FAMB; further dividing overlapping features based on the attribution division mechanism of fuzzy dependency; and finally determining a heterogeneous feature selection algorithm with group structure mining to screen and optimize the feature group; The diabetic nephropathy classification model is trained by using the divided DN group data set and the NDRD group data set until a trained diabetic nephropathy classification model is obtained.

2. The method for constructing a diabetic nephropathy classification model according to claim 1, characterized in that: The method further comprises: The constructed diabetic nephropathy classification model was compared with other feature selection algorithm models; among which, other feature selection algorithm models at least included DRGS, FastCFS, FCBF, IWFS, MRI, MRMI and N-MRMCR-MI algorithms; And the classification results are determined by multiple classifiers; wherein the classifiers include at least Classification And Regression Tree, Bayes, K-Nearest Neighbor, and Support Vector Machine classifiers; The evaluation results of each model were obtained based on classification accuracy and average reduction rate.

3. The method for constructing a diabetic nephropathy classification model according to claim 2, characterized in that: The model comparison and evaluation was realized through Matlab software, and different algorithm models were generated to obtain the relationship diagram between the number of selected features and classification accuracy.

4. The method for constructing a diabetic nephropathy classification model according to claim 1, characterized in that: Before training the diabetic nephropathy classification model using the divided DN group data set and NDRD group data set, the method further includes: Split the dataset into training / testing datasets using cross-validation.

5. The method for constructing a diabetic nephropathy classification model according to claim 1, characterized in that: According to the target diabetic nephropathy data set, the diabetic nephropathy detection indicators are analyzed and determined to include: The patient's ID, physical parameters, diagnosis results, prealbumin, creatine kinase, urea, apolipoprotein A1, low-density lipoprotein cholesterol, albumin / globulin, glomerular filtration rate, alkaline phosphatase, apolipoprotein B, amylase, α-hydroxybutyrate dehydrogenase, V-glutamyl transpeptidase, indirect bilirubin, total calcium, potassium, and chloride.

6. A device for constructing a diabetic nephropathy classification model based on a fuzzy decision system, characterized in that: The device comprises: The preprocessing module is used to analyze and determine the diabetic nephropathy detection indicators according to the target diabetic nephropathy data set; import the data corresponding to each diabetic nephropathy detection indicator into SPSS software and preprocess the data; and use the adjacent point average method to process the missing values ​​in the data set; according to the preprocessed biopsy data, the data is divided into a DN group data set and a NDRD group data set; A construction module is used to construct a diabetic nephropathy classification model; wherein the diabetic nephropathy classification model specifically includes introducing a Markov blanket algorithm and fuzzy rough set approximation and fuzzy uncertainty measurement, determining a feature group generation strategy based on FAMB; further dividing overlapping features based on a fuzzy dependency attribution division mechanism; and finally determining a heterogeneous feature selection algorithm with group structure mining to screen and optimize feature groups; The training module is used to train the diabetic nephropathy classification model through the divided DN group data set and the NDRD group data set until a trained diabetic nephropathy classification model is obtained.

7. The device for constructing a diabetic nephropathy classification model according to claim 6, characterized in that: The device also includes: A comparison module is used to compare the constructed diabetic nephropathy classification model with other feature selection algorithm models; wherein the other feature selection algorithm models include at least DRGS, FastCFS, FCBF, IWFS, MRI, MRMI and N-MRMCR-MI algorithms; and to determine the classification results through multiple classifiers; wherein the classifiers include at least ClassificationAndRegressionTree, Bayes, K-Nearest Neighbor and SupportVectorMachine classifiers; the evaluation results of each model are obtained based on classification accuracy and average reduction rate.

8. The device for constructing a diabetic nephropathy classification model according to claim 7, characterized in that: The model comparison and evaluation was realized through Matlab software, and different algorithm models were generated to obtain the relationship diagram between the number of selected features and classification accuracy.

9. The device for constructing a diabetic nephropathy classification model according to claim 6, characterized in that: Before training the diabetic nephropathy classification model using the divided DN group data set and NDRD group data set, the method further includes: Split the dataset into training / testing datasets using cross-validation.

10. The device for constructing a diabetic nephropathy classification model according to claim 6, characterized in that: According to the target diabetic nephropathy data set, the diabetic nephropathy detection indicators are analyzed and determined to include: The patient's ID, physical parameters, diagnosis results, prealbumin, creatine kinase, urea, apolipoprotein A1, low-density lipoprotein cholesterol, albumin / globulin, glomerular filtration rate, alkaline phosphatase, apolipoprotein B, amylase, α-hydroxybutyrate dehydrogenase, V-glutamyl transpeptidase, indirect bilirubin, total calcium, potassium, and chloride.

Images