Kidney prognosis risk prediction method and system for patient diagnosed with type 2 diabetic nephropathy through pathology

Abstract

The invention provides a kidney prognosis risk prediction method and system for a patient diagnosed as type 2 diabetic nephropathy through pathology, and relates to the technical field of prognosis risk prediction.The method comprises the steps that clinical data of the T2DN patient meeting the screening standard are obtained; screening the clinical data through an LASSO regression method so as to screen out key clinical indexes related to kidney prognosis; based on the key clinical indexes, constructing a risk prediction model by using a Cox proportional risk model; performing performance analysis on the risk prediction model by using clinical data of the training queue and the verification queue, and performing internal verification on the risk prediction model by using a Bootstrap method to ensure the robustness and reliability of the model; and inputting clinical data of a to-be-tested patient into the risk prediction model to obtain a prediction result. By screening key clinical indexes, an individualized risk prediction model is constructed, and a scientific basis is provided for clinical early intervention.

Description

Technical Field

[0001] The present invention relates to the technical field of prognosis risk prediction, and in particular to a method and system for predicting kidney prognosis risk of patients with type 2 diabetic nephropathy diagnosed by pathology. Background Art

[0002] Type 2 diabetic nephropathy (T2DN) is one of the most common chronic complications in diabetic patients and the main cause of end-stage renal disease (ESRD). Early identification of the renal prognosis risk of T2DN patients is of great significance for delaying disease progression and optimizing treatment strategies. However, there is currently a lack of effective tools in clinical practice to predict the risk of progression of T2DN patients from CKD stages 1-4 to ESRD. Summary of the invention

[0003] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a method and system for predicting renal prognosis risk in patients pathologically diagnosed with type 2 diabetic nephropathy, by screening key clinical indicators and constructing an individualized risk prediction model, so as to provide a scientific basis for early clinical intervention.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A method for predicting renal prognosis risk in patients with pathologically confirmed type 2 diabetic nephropathy, comprising:

[0006] Obtain clinical data of T2DN patients who meet the screening criteria;

[0007] The clinical data are screened by LASSO regression method to screen out key clinical indicators related to renal prognosis;

[0008] Based on the key clinical indicators, a risk prediction model was constructed using the Cox proportional hazards model;

[0009] The performance of the risk prediction model was analyzed using clinical data from the training cohort and the validation cohort, and the risk prediction model was internally validated using the Bootstrap method to ensure the robustness and reliability of the model;

[0010] The clinical data of the patient to be tested is input into the risk prediction model to obtain a prediction result.

[0011] Preferably, the clinical data include demographic characteristics, biochemical indicators and pathological results.

[0012] Preferably, the key clinical indicators include: urine total protein / urine creatinine, albumin, baseline estimated glomerular filtration rate and urine protein-creatinine ratio, age, body mass index, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol, blood creatinine, blood uric acid, glycated hemoglobin and hemoglobin.

[0013] Preferably, the output results of the risk prediction model are presented through a nomogram; the nomogram provides an individualized risk scoring tool by integrating the regression coefficients of each of the key clinical indicators; the nomogram is used to quantitatively predict the risk of a patient progressing to ESRD within 3 and 5 years.

[0014] Preferably, the performance analysis indicators include: ROC curve, calibration curve and decision curve.

[0015] Preferably, the discrimination of the risk prediction model is evaluated by the AUC value of the ROC curve, and the AUC value is greater than 0.85 in both the training cohort and the validation cohort.

[0016] Preferably, the calibration of the risk prediction model is analyzed by a calibration curve to show good consistency between the predicted probability and the actual observed results.

[0017] Preferably, the clinical utility of the risk prediction model is analyzed by the decision curve to show that the net benefit of the model at different risk thresholds is superior to the "full treatment" or "no treatment" strategy.

[0018] Preferably, the risk prediction model is internally validated using the Bootstrap method, including:

[0019] The Bootstrap method was used to repeat the sampling 500 times for internal validation and obtain the validation results.

[0020] A renal prognosis risk prediction system for patients with type 2 diabetic nephropathy diagnosed by pathology, comprising:

[0021] A data acquisition module, used to obtain clinical data of T2DN patients who meet the screening criteria;

[0022] A data screening module, used for screening the clinical data by LASSO regression method to screen out key clinical indicators related to kidney prognosis;

[0023] A model building module, used to build a risk prediction model based on the key clinical indicators using a Cox proportional hazard model;

[0024] A performance analysis module, used to perform performance analysis on the risk prediction model using clinical data from a training cohort and a validation cohort, and to internally validate the risk prediction model using a Bootstrap method to ensure the robustness and reliability of the model;

[0025] The prediction module is used to input the clinical data of the patient to be tested into the risk prediction model to obtain a prediction result.

[0026] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0027] The present invention provides a method and system for predicting the renal prognosis risk of patients with type 2 diabetic nephropathy diagnosed by pathology, the method comprising: obtaining clinical data of T2DN patients meeting the screening criteria; screening the clinical data by LASSO regression method to screen out key clinical indicators related to renal prognosis; constructing a risk prediction model based on the key clinical indicators using the Cox proportional hazard model; using the clinical data of the training cohort and the validation cohort to perform performance analysis on the risk prediction model, and using the Bootstrap method to perform internal verification on the risk prediction model to ensure the robustness and reliability of the model; inputting the clinical data of the patient to be tested into the risk prediction model to obtain a prediction result. The present invention constructs an individualized risk prediction model by screening key clinical indicators, providing a scientific basis for early clinical intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0029] Figure 1 A flow chart of a method provided by an embodiment of the present invention;

[0030] Figure 2 A schematic diagram of the Lasso coefficients of 18 clinical indicators provided in an embodiment of the present invention;

[0031] Figure 3 A schematic diagram of log(λ) and partial likelihood deviation provided by an embodiment of the present invention;

[0032] Figure 4 A schematic diagram of the three-year risk of predicting progression from CKD stages 1-4 to end-stage renal disease (ESRD) provided in an embodiment of the present invention;

[0033] Figure 5A schematic diagram of the 5-year risk of predicting progression from CKD stages 1-4 to end-stage renal disease (ESRD) provided in an embodiment of the present invention;

[0034] Figure 6 A first schematic diagram of AUC in a scalar graph of a 3-year training cohort provided by an embodiment of the present invention;

[0035] Figure 7 A first schematic diagram of AUC in a scalar graph of a 5-year training cohort provided by an embodiment of the present invention;

[0036] Figure 8 A second schematic diagram of AUC in a scalar graph of a 3-year training cohort provided by an embodiment of the present invention;

[0037] Fig. 9 A second schematic diagram of AUC in a scalar graph of a 5-year training cohort provided by an embodiment of the present invention;

[0038] Fig.10 A schematic diagram showing the predicted probability and the actual observed result of a calibration histogram of the probability of end-stage renal disease (ESRD) occurring in a training set provided by an embodiment of the present invention;

[0039] Fig.11 A schematic diagram showing a calibration histogram of the probability of end-stage renal disease (ESRD) occurring in a validation set provided by an embodiment of the present invention, showing predicted probabilities and actual observed results;

[0040] Fig.12 A schematic diagram of a decision curve analysis of the 3-year diabetes risk in a training cohort using the LASSO model provided in an embodiment of the present invention;

[0041] Fig.13 A schematic diagram of a decision curve analysis of the 3-year diabetes risk in a validation cohort using the LASSO model provided in an embodiment of the present invention;

[0042] Fig.14 A schematic diagram of a survival curve of renal survival rate (not entering ESRD) of patients in a training cohort provided in an embodiment of the present invention;

[0043] Fig.15 A schematic diagram of a survival curve of renal survival rate (not entering ESRD) of patients in a validation cohort provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The purpose of the present invention is to provide a method and system for predicting the renal prognosis risk of patients with type 2 diabetic nephropathy diagnosed by pathology, by screening key clinical indicators and constructing an individualized risk prediction model, so as to provide a scientific basis for early clinical intervention.

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] Figure 1 A flow chart of a method provided by an embodiment of the present invention, such as Figure 1 As shown, the present invention provides a method for predicting the renal prognosis risk of patients with type 2 diabetic nephropathy diagnosed by pathology, comprising:

[0048] Step 100: Acquire clinical data of T2DN patients who meet the screening criteria; illustratively, the T2DN patients in this embodiment are patients who have undergone renal puncture diagnosis (ie, biopsy).

[0049] Step 200: screening clinical data by LASSO regression method to screen out key clinical indicators related to renal prognosis;

[0050] Step 300: constructing a risk prediction model using a Cox proportional hazard model based on key clinical indicators;

[0051] Step 400: using the clinical data of the training cohort and the validation cohort to perform performance analysis on the risk prediction model, and using the Bootstrap method to perform internal validation on the risk prediction model to ensure the robustness and reliability of the model;

[0052] Step 500: Input the clinical data of the patient to be tested into the risk prediction model to obtain a prediction result.

[0053] Preferably, the clinical data include demographic characteristics, biochemical indicators and pathological results.

[0054] Specifically, the clinical data of the training cohort and the validation cohort of this embodiment are shown in Table 1:

[0055] Table 1 Clinical data of training cohort and validation cohort

[0056]

[0057]

[0058]

[0059] Optionally, after acquiring the clinical data, this embodiment further preprocesses the clinical data, and the preprocessing process includes:

[0060] Step 1011: Missing value processing, using multiple imputation techniques to process missing values.

[0061] Step 1012: Outlier detection and processing, using a machine learning algorithm (such as isolation forest) to detect outliers.

[0062] Step 1013: Data standardization and normalization, standardizing or normalizing the numerical features.

[0063] Step 1014: Text classification: Classify the clinical text data to extract useful information.

[0064] Specifically, the method for classifying clinical text data in this embodiment is as follows:

[0065] Step 10141: input the clinical data into the text context feature extraction layer to extract context feature information;

[0066] Step 10142: inputting the clinical data into a global feature extraction layer to extract global feature information;

[0067] Step 10143: Fusing the context feature information and the global feature information to obtain a fused text feature;

[0068] Step 10144: construct a loss function using fused text features;

[0069] Step 10145: Continuously optimize the loss function to obtain a text classification model;

[0070] Step 10146: Use the text classification model to complete text classification.

[0071] Furthermore, in this embodiment, the clinical data is input into the text context feature extraction layer to extract context feature information, including:

[0072] Use the pre-trained language model to extract the initial feature information of the clinical data;

[0073] Inputting the initial feature information into the forward gated recurrent unit and the reverse gated recurrent unit;

[0074] The outputs of the forward gated recurrent unit and the reverse gated recurrent unit are concatenated to obtain contextual feature information; the formula of the contextual feature information is: Among them, H t ={H1,H2,…H l} represents the initial feature information input at time t, represents the output of the positive gated recurrent unit at time t, Represents the output of the reverse gated recurrent unit at time t, GRU represents the gated recurrent unit, and G represents the contextual relationship feature information.

[0075] Furthermore, in this embodiment, the clinical data is input into the global feature extraction layer to extract global feature information, including:

[0076] The initial feature information of the text to be classified is input into the convolution layer and the pooling layer in sequence to obtain the global feature information; wherein, the global feature information extraction formula is:

[0077] c i =f(ω·H+b)

[0078]

[0079] In the formula, f is the activation function, ω is the convolution kernel, h is the convolution kernel size, b is the bias value, c is i is the feature vector extracted by the i-th convolutional layer, It represents the value of the features extracted by three different convolution kernels after the maximum pooling layer, and C represents the extracted global feature information.

[0080] Furthermore, the loss function of this embodiment is expressed as:

[0081]

[0082] in, represents the true category of the clinical data text, m represents the number of total text categories, λ represents the regularization parameter, and y k Indicates the prediction result.

[0083] For example, after defining the loss function, the next step is to minimize the loss function through an optimization algorithm to train the text classification model. Common optimization algorithms include stochastic gradient descent (SGD), Adam, etc. In specific implementation, the model parameters are first initialized, and then in each training iteration, the loss value of the current model is calculated, and the model parameters are updated according to the gradient of the loss value. The backpropagation algorithm can be used to calculate the gradient, and the parameters are updated through the optimization algorithm. During the training process, hyperparameters such as learning rate and batch size can be set, and the validation set can be used to monitor the model performance to prevent overfitting. Through multiple iterations, the loss function is gradually optimized until the model converges or the preset number of training rounds is reached.

[0084] Optionally, once the model training is completed, the present embodiment uses the trained text classification model to classify new clinical text data. First, the text data to be classified is subjected to the same preprocessing, including cleaning, word segmentation and feature extraction. Then, the processed text features are input into the trained model, and the model outputs the predicted probability of each category. According to the set threshold (usually 0.5 or adjusted according to specific needs), the predicted probability is converted into a specific category label. Finally, the classification result is combined with the original text data to form the final classification output to obtain the classified clinical data.

[0085] Preferably, the key clinical indicators include: urine total protein / urine creatinine, albumin, baseline estimated glomerular filtration rate and urine protein-creatinine ratio, age, body mass index, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol, blood creatinine, blood uric acid, glycated hemoglobin and hemoglobin.

[0086] like Figure 2 and Figure 3 As shown, the diagram shows the process of selecting prediction factors through the Lasso regression model in this embodiment. Figure 2 is the Lasso coefficient of 18 clinical indicators. Figure 3 The log(λ) and partial likelihood deviation are shown; the dashed line indicates the minimum log(λ), representing the optimal number of predictors. Specifically, the key clinical indicators of this embodiment are shown in Table 2.

[0087] Table 2 Screening of risk factors using stepwise Coxregressionmodel

[0088]

[0089] like Figure 4 and Figure 5As shown, the corresponding scalar graphs constructed based on albumin (ALB), urine protein to creatinine ratio (PCR) and baseline estimated glomerular filtration rate (eGFR) in this embodiment provide a quantitative and simple tool for predicting the 3-year (3A) and 5-year (5A) risk of progression from CKD stage 1-4 to end-stage renal disease (ESRD). Exemplarily, the expression of the risk prediction model Model in this embodiment is:

[0090] Model = -0.05804*blood albumin (g / L) -0.03257*baseline estimated glomerular filtration rate (ml / min) +0.00009*random urine total protein / urine creatinine (mg / g).

[0091] Preferably, the output results of the risk prediction model are presented through a nomogram; the nomogram provides an individualized risk scoring tool by integrating the regression coefficients of each of the key clinical indicators; the nomogram is used to quantitatively predict the risk of a patient progressing to ESRD within 3 and 5 years.

[0092] Preferably, the performance analysis indicators include: ROC curve, calibration curve and decision curve.

[0093] Preferably, the discrimination of the risk prediction model is evaluated by the AUC value of the ROC curve, and the AUC value is greater than 0.85 in both the training cohort and the validation cohort.

[0094] Specifically, Figures 6 to 9 As shown, the time in the figure is the number of days, and the AUC of the scalar graph is 0.84588 (95% CI, (0.7665, 0.8616)) Figure 6 , 3-year training cohort) and 0.85868 (95% CI, (0.7665, 0.8616)) ( Figure 7 , 5-year training cohort). The AUC of this scalar map was 0.92906 (95% CI, (0.8102, 0.9309)) in the 3-year validation cohort ( Figure 8 ) and 0.86984 (95% CI, (0.8102, 0.9309)) in the 5-year validation cohort ( Fig. 9 ).

[0095] Preferably, the calibration of the risk prediction model is analyzed by a calibration curve to show good consistency between the predicted probability and the actual observed results.

[0096] Specifically, Fig.10 and Fig.11 As shown, the calibration histograms of the probability of developing end-stage renal disease (ESRD) in the training and validation sets show good agreement between the predicted probabilities and the actual observed results.

[0097] Preferably, the clinical utility of the risk prediction model is analyzed by the decision curve to show that the net benefit of the model at different risk thresholds is superior to the "full treatment" or "no treatment" strategy.

[0098] Preferably, the risk prediction model is internally validated using the Bootstrap method, including:

[0099] The Bootstrap method was used to repeat the sampling 500 times for internal validation and obtain the validation results.

[0100] Specifically, the LASSO model is used in the training queue ( Fig.12 ) and the validation queue ( Fig.13 ) Decision curve analysis of 3-year diabetes risk in a scalar plot. The black line represents the net benefit when no participant develops end-stage renal disease (ESRD) and dialysis, while the light gray line represents the net benefit when all participants are assumed to develop ESRD and dialysis. The area between the “no treatment line” (black line) and the “full treatment line” (light gray line) in the model curve represents the clinical utility of the model. The farther the model curve is from the black and light gray lines, the better the clinical application of the scalar plot.

[0101] In addition, if Fig.14 and Fig.15 As shown, this embodiment divides patients into a low-risk group and a high-risk group according to the predicted risk factors, and the figure shows the survival curve of the renal survival rate (not entering ESRD) of patients in the training cohort and the validation cohort.

[0102] Corresponding to the above method, this embodiment also provides a renal prognosis risk prediction system for patients with type 2 diabetic nephropathy diagnosed by pathology, comprising:

[0103] A data acquisition module, used to obtain clinical data of T2DN patients who meet the screening criteria;

[0104] A data screening module, used for screening the clinical data by LASSO regression method to screen out key clinical indicators related to kidney prognosis;

[0105] A model building module, used to build a risk prediction model based on the key clinical indicators using a Cox proportional hazard model;

[0106] A performance analysis module, used to perform performance analysis on the risk prediction model using clinical data from a training cohort and a validation cohort, and to internally validate the risk prediction model using a Bootstrap method to ensure the robustness and reliability of the model;

[0107] The prediction module is used to input the clinical data of the patient to be tested into the risk prediction model to obtain a prediction result.

[0108] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0109] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for predicting renal prognosis risk in patients with type 2 diabetic nephropathy diagnosed by pathology, characterized in that: include: Obtain clinical data of T2DN patients who meet the screening criteria; The clinical data are screened by LASSO regression method to screen out key clinical indicators related to renal prognosis; Based on the key clinical indicators, a risk prediction model was constructed using the Cox proportional hazards model; The performance of the risk prediction model was analyzed using clinical data from the training cohort and the validation cohort, and the risk prediction model was internally validated using the Bootstrap method to ensure the robustness and reliability of the model; The clinical data of the patient to be tested is input into the risk prediction model to obtain a prediction result.

2. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy pathologically confirmed according to claim 1, characterized in that: The clinical data included demographic characteristics, biochemical indicators and pathological results.

3. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy pathologically confirmed according to claim 1, characterized in that: The key clinical indicators include: urine total protein / urine creatinine, albumin, baseline estimated glomerular filtration rate and urine protein-creatinine ratio, age, body mass index, systolic blood pressure, diastolic blood pressure, triglycerides, total cholesterol, blood creatinine, blood uric acid, glycated hemoglobin and hemoglobin.

4. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy pathologically confirmed according to claim 1, characterized in that: The output results of the risk prediction model are presented through a nomogram; the nomogram provides an individualized risk scoring tool by integrating the regression coefficients of each of the key clinical indicators; the nomogram is used to quantitatively predict the risk of a patient progressing to ESRD within 3 and 5 years.

5. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy pathologically confirmed according to claim 1, characterized in that: The performance analysis indicators include: ROC curve, calibration curve and decision curve.

6. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy pathologically confirmed according to claim 5, characterized in that: The discrimination of the risk prediction model was evaluated by the AUC value of the ROC curve, and the AUC value was greater than 0.85 in both the training cohort and the validation cohort.

7. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy pathologically confirmed according to claim 5, characterized in that: The calibration of the risk prediction model was analyzed by calibration curve to show good consistency between the predicted probability and the actual observed results.

8. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy diagnosed pathologically according to claim 5, characterized in that: The clinical utility of the risk prediction model is analyzed by the decision curve to show that the net benefit of the model is superior to the "full treatment" or "no treatment" strategy at different risk thresholds.

9. The method for predicting renal prognosis risk of patients with type 2 diabetic nephropathy diagnosed pathologically according to claim 1, characterized in that: The risk prediction model was internally validated using the Bootstrap method, including: The Bootstrap method was used to repeat the sampling 500 times for internal validation and obtain the validation results.

10. A renal prognosis risk prediction system for patients with type 2 diabetic nephropathy diagnosed by pathology, characterized in that: include: A data acquisition module, used to obtain clinical data of T2DN patients who meet the screening criteria; A data screening module, used for screening the clinical data by LASSO regression method to screen out key clinical indicators related to kidney prognosis; A model building module, used to build a risk prediction model based on the key clinical indicators using a Cox proportional hazard model; A performance analysis module, used to perform performance analysis on the risk prediction model using clinical data from a training cohort and a validation cohort, and to internally validate the risk prediction model using a Bootstrap method to ensure the robustness and reliability of the model; The prediction module is used to input the clinical data of the patient to be tested into the risk prediction model to obtain a prediction result.

Images