System and device for predicting the prognosis of diabetic nephropathy based on weighted quantile and regression

Abstract

The present invention discloses a system and device for predicting the prognosis of diabetic nephropathy based on weighted quantiles and regression, belonging to the technical field of computer systems based on specific computing models, the system includes an acquisition module, a weighting module and a prediction module, the acquisition module is used to collect clinical indicators and retinal vascular geometric parameters of patients; the weighting module is used to construct a weighted index through the retinal vascular geometric parameters based on weighted quantiles and regression; the prediction module is used to analyze the clinical indicators and weighted index based on the prediction model to obtain the incidence of prognostic events. The weighted index represents the mixed effect level of each retinal vascular geometric parameter, and the weighted index reflects the overall effect of the retinal vascular geometric parameters on the identification of diabetic nephropathy. It performs well in characterizing the mixed effect and can improve the prediction performance. The diagnostic performance of the prediction model is good and can be applied to clinical practice.

Description

Technical Field

[0001] The present invention relates to the technical field of computer systems based on specific computing models, and in particular to a system and device for predicting the prognosis of diabetic nephropathy based on weighted quantile and regression. Background Art

[0002] Diabetic nephropathy is the most common cause of end-stage renal disease. Patients with end-stage renal disease must undergo dialysis or kidney transplantation to prolong survival, placing a heavy burden on patients and society. Early detection and improved management of patients with diabetic nephropathy can slow the progression, reduce its complications, and improve outcomes.

[0003] A large number of previous studies predicting the prognosis of diabetic nephropathy have shown that cystatin C, albumin, hemoglobin, 24-hour urine total protein, eGFR, duration of diabetes, fundus lesions, BMI, hypertension grade, chronic kidney disease stage, glomerular hyalinosis, and increased extracellular cells in capillaries are all factors that affect the prognosis of diabetic nephropathy. The prognosis of diabetic nephropathy refers to the prognosis of the development of diabetic nephropathy, including survival, death, or renal failure. The purpose of studying prognosis is to understand the laws of the disease development process, which is one of the driving forces of medical progress. Therefore, the association between clinical indicators of diabetic nephropathy and retinal microvasculature and prognosis is an important research direction. Summary of the invention

[0004] In response to the above-mentioned technical problems existing in the prior art, the present invention provides a system and device for predicting the prognosis of diabetic nephropathy based on weighted quantile and regression, predicting the prognosis of diabetic nephropathy based on clinical indicators and retinal microvessels, and providing data support for clinical treatment.

[0005] The present invention discloses a system for predicting the prognosis of diabetic nephropathy based on weighted quantile and regression, comprising an acquisition module, a weighting module and a prediction module, wherein the acquisition module is used to acquire clinical indicators and retinal vascular geometric parameters of patients; the weighting module is used to construct a weighted index through the retinal vascular geometric parameters based on weighted quantile and regression; the prediction module is used to analyze the clinical indicators and the weighted index based on a prediction model to obtain the incidence of prognostic events.

[0006] Preferably, the clinical indicators are selected from the following characteristics:

[0007] Cystatin C, 24-hour urine protein, albumin, hemoglobin, glomerular filtration rate, diastolic blood pressure, systolic blood pressure, diabetic retinopathy, triglycerides, low-density lipoprotein, glycosylated hemoglobin, body mass index, sex, high-density lipoprotein, uric acid, age, drinking history, smoking history, and duration of diabetes.

[0008] Preferably, the retinal vascular geometric parameters are selected from the following features:

[0009] Blood vessel fractal dimension variable b, arterial fractal dimension variable k, venous tortuosity, arterial tortuosity, venous fractal dimension variable k, venous fractal dimension variable b, arterial diameter, arterial tortuosity, blood vessel tortuosity, blood vessel fractal dimension variable k, blood vessel diameter, venous diameter, tortuosity density, branching angle, venous tortuosity density, and arterial fractal dimension variable b.

[0010] Preferably, the weighted index is calculated by:

[0011] WQS=B0+B1*Fb+B2*Fka+B3*DMv+B4*DMa+B5*Fkv+B6*Fbv+B7*Da+B8*TDa+B9*DM+B 10 *Fk+B 11 *D+B 12 *Dv+B 13 *TD+B 14 *B+B 15 *TDv+B 16 *Fba

[0012] Among them, WQS represents the weighting parameter, B0 is a constant, B1-B 16 is the weight coefficient, Fb represents the vascular fractal dimension variable b, Fka represents the arterial fractal dimension variable k, DMv represents the venous tortuosity, DMa represents the arterial tortuosity, Fkv represents the venous fractal dimension variable k, Fbv represents the venous fractal dimension variable b, Da is the arterial diameter, TDa is the arterial bending density, DM is the vascular tortuosity, Fk is the vascular fractal dimension variable k, D is the vascular diameter, Dv is the venous diameter, TD is the bending density, B is the branching angle, TDv is the venous bending density, and Fba is the arterial fractal dimension variable b.

[0013] Preferably, the specific weighted index is calculated as follows:

[0014] WQS=Fb*0.407+Fka*0.107+DMv*0.085+DMa*0.080+Fkv*0.062+Fbv*0.059+Da*0.056+ TDa*0.054+DM*0.033+Fk*0.019+D*0.012+Dv*0.010+TD*0.008+B*0.004+TDv*0.001.

[0015] Preferably, the system further comprises a training module, and the training module is used to train the prediction model.

[0016] Preferably, the method for training the prediction model comprises:

[0017] Screen and collect retinal vascular geometric parameters;

[0018] A weighted index was constructed using retinal vascular geometric parameters;

[0019] Screen and collect clinical indicators;

[0020] Constructing a training set and a test set according to the weighted index and clinical indicators;

[0021] Based on the machine learning method, a prediction model is obtained through the training set.

[0022] Preferably, the training set includes the prognostic events of the patients at the follow-up time points, and the prognostic events include: survival, and death / renal failure.

[0023] Preferably, the machine learning method is selected from: COX-Ridge, COX-Lasso, COX-Is_Tree, COX-Cph_Tree, COX-Elastic_net, COX-RF and COX-least_squares.

[0024] The present invention also provides a device, comprising an input module, a processing module and an output module.

[0025] The input module is used to obtain the patient's clinical indicators and retinal vascular geometric parameters;

[0026] The processing module is used to construct a weighted index through the retinal vascular geometric parameters based on weighted quantiles and regression; based on the prediction model, the clinical indicators and weighted index are analyzed to obtain the incidence of prognostic events;

[0027] The output module is used to output the occurrence rate.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows: the weighted index represents the level of mixed effect of various retinal vascular geometric parameters, and reflects the overall effect of retinal vascular geometric parameters on the identification of diabetic nephropathy through the weighted index. It performs well in characterizing the mixed effect and can improve the prediction performance. The diagnostic performance of the prediction model is good and can be applied in clinical practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a significant level analysis diagram of each feature of the prediction model;

[0030] Figure 2A is the AUC curve of the COX-Ridge prediction model;

[0031] Figure 2Bis the AUC curve of the COX-RF prediction model;

[0032] Figure 2C is the AUC curve of the COX-Lasso prediction model;

[0033] Figure 2D is the AUC curve of the COX-Is_Tree prediction model;

[0034] Figure 2E is the AUC curve of the COX-Cph_Tree prediction model;

[0035] Figure 2F is the AUC curve of the COX-Elastic_net prediction model;

[0036] Figure 3A This is the AUC curve of the COX-RF prediction model based on clinical indicators;

[0037] Figure 3B This is the AUC curve of the COX-RF prediction model based on retinal vascular geometric parameters;

[0038] Figure 4 is the survival curve;

[0039] Figure 5 It is a system logic block diagram of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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.

[0041] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0042] The first aspect of the present invention provides a method for predicting the prognosis of diabetic nephropathy based on weighted quantile and regression, comprising the following steps:

[0043] Step 101: Collect the patient's clinical indicators and retinal vascular geometric parameters.

[0044] Step 102: constructing a weighted index (WQS index) through the retinal vascular geometric parameters based on a weighted quantile sum regression (WQS) method.

[0045] Step 103: Based on the prediction model, the clinical indicators and weighted indexes are analyzed to obtain the incidence of prognostic events.

[0046] The prognostic events included survival and death / renal failure.

[0047] The weighted index represents the level of mixed effect of various retinal vascular geometric parameters. The weighted index is used to explore the overall effect of retinal vascular geometric parameters on the identification of diabetic nephropathy. It performs well in characterizing the mixed effect and can improve the prediction performance. The diagnostic performance of the prediction model is good and can be applied in clinical practice.

[0048] The method of predictive model training includes the following steps:

[0049] Step 201: Screening and collecting retinal blood vessel geometric parameters.

[0050] Among them, the retinal vascular geometric parameters are selected from: vascular diameter (D, average vascular diameter), arterial diameter (Da), venous diameter (Dv), vascular fractal dimension variable k (Fk), fractal dimension variable b (Fb), arterial fractal dimension variable k (Fka), arterial fractal dimension variable b (Fba), venous fractal dimension variable k (Fkv), venous fractal dimension variable b (Fbv), vascular tortuosity (DM), arterial tortuosity (DMa), venous tortuosity (DMv), curvature density (TD), arterial curvature density (TDa), venous curvature density (TDv), and branching angle (B).

[0051] Among them, the blood vessel diameter, fractal dimension, curvature, curvature density, and branch angle are calculated based on the features extracted from the eye image, and most of them are calculated using averages. The curvature is the ratio of the arc length of the blood vessel contour fitting curve divided by the chord length. The calculation and acquisition methods of the above geometric parameters are prior art and will not be repeated in the present invention.

[0052] Step 202: construct a weighted index using retinal vascular geometric parameters.

[0053] The weighted index (WQS index), which is an empirically weighted sum of the individual parameters, was calculated using the R package gWQS version 3.0.5. The WQS index (ranging from 0 to 1) represents the level of mixed effects of the parameters, with the parameters of interest being weighted by non-negligible weights. Figure 1 The weights of various retinal vascular geometric parameters are shown. The calculation formula of the WQS index is:

[0054] WQS=B0+B1*Fb+B2*Fka+B3*DMv+B4*DMa+B5*Fkv+B6*Fbv+B7*Da+B8*TDa+B9*DM+B 10 *Fk+B 11*D+B 12 *Dv+B 13 *TD+B 14 *B+B 15 *TDv+B 16 *Fba

[0055] Among them, WQS represents the weighting parameter, B0 is a constant, B1-B 16 is the weight coefficient, Fb represents the vascular fractal dimension variable b, Fka represents the arterial fractal dimension variable k, DMv represents the venous tortuosity, DMa represents the arterial tortuosity, Fkv represents the venous fractal dimension variable k, Fbv represents the venous fractal dimension variable b, Da is the arterial diameter, TDa is the arterial bending density, DM is the vascular tortuosity, Fk is the vascular fractal dimension variable k, D is the vascular diameter, Dv is the venous diameter, TD is the bending density, B is the branching angle, TDv is the venous bending density, and Fba is the arterial fractal dimension variable b.

[0056] In a specific training, the calculation formula of WQS index is:

[0057] WQS=Fb*0.407+Fka*0.107+DMv*0.085+DMa*0.080+Fkv*0.062+Fbv*0.059+Da*0.056+TD a*0.054+DM*0.033+Fk*0.019+D*0.012+Dv*0.010+TD*0.008+B*0.004+TDv*0.001+Fba*0

[0058] More specifically, parameters with coefficients less than 0.001 may be deleted from the calculation formula.

[0059] Step 203: Screen and collect clinical indicators.

[0060] In terms of clinical features of diagnosis, eleven clinical features / indicators were selected based on COX univariate regression analysis, including: cystatin C (Cystatin_C), 24-hour urine protein (Urine_protein_24), albumin ALB, hemoglobin Hb, glomerular filtration rate eGFR, diastolic blood pressure DBP, systolic blood pressure SBP, diabetic retinopathy DR, triglycerides TG, weighted index WQS, low-density lipoprotein LDL, glycosylated hemoglobin HbA1c, body mass index BMI, sex Sex, high-density lipoprotein HDL, uric acid UA, age Age, drinking history Drink, smoking history Smoke, and diabetic course Diabetic_courses. The significance level P of each feature is shown in Table 1.

[0061] Table 1

[0062]

[0063] Step 204: Construct a training set and a test set. The median follow-up time of the training set and the test set was 36.2 months and 40.0 months, respectively, and the incidence of endpoint events was 45.9% and 35.7%, respectively. See Table 2 for the data distribution of the training set and the test set.

[0064] Table 2

[0065]

[0066] Step 205: Prediction model training.

[0067] Based on six COX proportional hazard models, prediction models were trained to predict the prognosis of DN. Figure 2A-2F As shown, the performance of each prediction model was evaluated by AUC, which was comparable to COX-Ridge ( Figure 2A )、COX-Lasso( Figure 2C )、COX-Is_Tree( Figure 2D )、COX-Cph_Tree( Figure 2E ) and COX-Elastic_net ( Figure 2F ) compared to COX-RF ( Figure 2B ) showed the best performance, with an AUC of 0.88. COX-RF was selected to train clinical indicators and retinal vascular geometry parameters respectively. The test results are shown in Figure 3A and Figure 3B , AUC were lower than those of the COX-RF prediction model. It can be seen that the prediction performance can be improved by mixing the weighted index of clinical indicators. Figure 4 The KM survival curve of the test set is shown. The present invention can also train a prediction model based on COX-least_squares.

[0068] A second aspect of the present invention provides a system for implementing the above method, such as Figure 5 , including acquisition module 1, weighting module 2 and prediction module 3,

[0069] The acquisition module 1 is used to acquire the patient's clinical indicators and retinal vascular geometric parameters;

[0070] The weighting module 2 is used to construct a weighted index through the retinal vascular geometric parameters based on weighted quantiles and regression;

[0071] The prediction module 3 is used to analyze the clinical indicators and weighted indexes based on the prediction model to obtain the incidence of prognostic events.

[0072] The system further comprises a training module 4, which is used for training the prediction model.

[0073] The third aspect of the present invention also provides a device for implementing the above method, comprising an input module, a processing module and an output module.

[0074] The input module is used to obtain the patient's clinical indicators and retinal vascular geometric parameters;

[0075] The processing module is used to construct a weighted index through the retinal vascular geometric parameters based on the weighted quantile and regression method; based on the prediction model, the clinical indicators and the weighted index are analyzed to obtain the incidence of prognostic events;

[0076] The output module is used to output the occurrence rate.

[0077] The present invention adopts a weighted quantile regression model to calculate mixed effects, fully considering inter-individual differences and internal correlations, and effectively integrating parameters such as retinal microvessel diameter, arteriovenous ratio, curvature, fractal dimension and vascular density.

[0078] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A system for predicting the prognosis of diabetic nephropathy based on weighted quantile and regression, characterized in that: Including acquisition module, weighting module and prediction module, The acquisition module is used to acquire clinical indicators and retinal vascular geometric parameters of the patient; The weighting module is used to construct a weighted index through the retinal vascular geometric parameters based on weighted quantiles and regression; The prediction module is used to analyze the clinical indicators and weighted indexes based on the prediction model to obtain the incidence of prognostic events; The weighted index is calculated by: WQS=B0+B1*Fb+B2*Fka+B3*DMv+B4*DMa+B5*Fkv+B6*Fbv+B7*Da+B8*TDa+B9*DM+B 10 *Fk+B 11 *D+B 12 *Dv+B 13 *TD+B 14 *B+B 15 *TDv+B 16 *Fba; Where WQS is a weighted index, B0 is a constant, B1-B 16 is the weight coefficient, Fb represents the vascular fractal dimension variable b, Fka represents the arterial fractal dimension variable k, DMv represents the venous tortuosity, DMa represents the arterial tortuosity, Fkv represents the venous fractal dimension variable k, Fbv represents the venous fractal dimension variable b, Da is the arterial diameter, TDa is the arterial bending density, DM is the vascular tortuosity, Fk is the vascular fractal dimension variable k, D is the vascular diameter, Dv is the venous diameter, TD is the bending density, B is the branching angle, TDv is the venous bending density, and Fba is the arterial fractal dimension variable b.

2. The system according to claim 1, characterized in that The clinical indicators are selected from the following characteristics: Cystatin C, 24-hour urine protein, albumin, hemoglobin, glomerular filtration rate, diastolic blood pressure, systolic blood pressure, diabetic retinopathy, triglycerides, low-density lipoprotein, glycosylated hemoglobin, body mass index, sex, high-density lipoprotein, uric acid, age, drinking history, smoking history, and duration of diabetes.

3. The system according to claim 1, characterized in that The retinal vascular geometry parameters are selected from the following features: Blood vessel fractal dimension variable b, arterial fractal dimension variable k, venous tortuosity, arterial tortuosity, venous fractal dimension variable k, venous fractal dimension variable b, arterial diameter, arterial tortuosity, vascular tortuosity, vascular fractal dimension variable k, vascular diameter, venous diameter, tortuosity, branching angle, venous tortuosity, and arterial fractal dimension variable b.

4. The system according to claim 1, characterized in that The specific weighted index is calculated as follows: WQS=Fb*0.407+Fka*0.107+DMv*0.085+DMa*0.080+Fkv*0.062+Fbv*0.059+Da*0.056+ TDa*0.054+DM*0.033+Fk*0.019+D*0.012+Dv*0.010+TD*0.008+B*0.004+TDv*0.

001.

5. The system according to claim 1, characterized in that It also includes a training module, which is used to train the prediction model.

6. The system according to claim 5, characterized in that The method of training the prediction model comprises: Screen and collect retinal vascular geometric parameters; A weighted index was constructed using retinal vascular geometric parameters; Screen and collect clinical indicators; Constructing a training set and a test set according to the weighted index and clinical indicators; Based on the machine learning method, a prediction model is obtained through the training set.

7. The system according to claim 6, characterized in that The training set includes the prognostic events of the patients at the follow-up time points, and the prognostic events include: survival, and death / renal failure.

8. The system according to claim 6, characterized in that The machine learning method is selected from: COX-Ridge, COX-Lasso, COX-Is_Tree, COX-Cph_Tree, COX-Elastic_net, COX-RF and COX-least_squares.

9. A device, characterized in that: It includes input module, processing module and output module. The input module is used to obtain the patient's clinical indicators and retinal vascular geometric parameters; The processing module is used to construct a weighted index through the retinal vascular geometric parameters based on weighted quantiles and regression; based on the prediction model, the clinical indicators and weighted index are analyzed to obtain the incidence of prognostic events; The output module is used to output the incidence rate; The weighted index is calculated by: WQS=B0+B1*Fb+B2*Fka+B3*DMv+B4*DMa+B5*Fkv+B6*Fbv+B7*Da+B8*TDa+B9*DM+B 10 *Fk+B 11 *D+B 12 *Dv+B 13 *TD+B 14 *B+B 15 *TDv+B 16 *Fba; Where WQS is a weighted index, B0 is a constant, B1-B 16 is the weight coefficient, Fb represents the vascular fractal dimension variable b, Fka represents the arterial fractal dimension variable k, DMv represents the venous tortuosity, DMa represents the arterial tortuosity, Fkv represents the venous fractal dimension variable k, Fbv represents the venous fractal dimension variable b, Da is the arterial diameter, TDa is the arterial bending density, DM is the vascular tortuosity, Fk is the vascular fractal dimension variable k, D is the vascular diameter, Dv is the venous diameter, TD is the bending density, B is the branching angle, TDv is the venous bending density, and Fba is the arterial fractal dimension variable b.

Images