Prediction model for interstitial fibrosis and cellular crescent body of chronic kidney disease and application of prediction model

Abstract

The invention discloses a chronic kidney disease interstitial fibrosis and cellular crescent body prediction model and application thereof. The prediction model comprises a kidney interstitial fibrosis prediction model and a cellular crescent body prediction model. The renal interstitial fibrosis prediction model is Logit (P1) =-0.035 * estimated glomerular filtration rate (ml / min / 1.73 m < 2 >) + 0.030 * albumin (g / L)-0.039 * hematocrit (%) + 0.036 * RBC distribution width SD (fL)-0.212 * CD4 + / CD8 + T lymphocyte ratio + 0.313; wherein P1 is a severe renal interstitial fibrosis probability value; the cell crescent body prediction model is Logit (P2) =-0.032 * age (age)-0.766 * sex-1.047 * high density lipoprotein (mmol / L)-0.011 * estimated glomerular filtration rate (ml / min / 1.73 m < 2 >)-0.233 * ln (CD3 + CD4 + T lymphocyte count) (piece / l) + 0.670 * ln (urine erythrocyte count) ( / l) + 1.530; wherein P2 is a cellular crescent body formation probability value. The prediction model provided by the invention greatly improves the prediction ability of renal interstitial fibrosis and cellular crescent body, and has significant clinical application value.

Description

Technical Field

[0001] The present invention belongs to the technical field of renal pathological prediction, and particularly relates to a prediction model for chronic kidney disease interstitial fibrosis and cellular crescents and its application. Background Art

[0002] Chronic kidney disease has become a major public health issue of global concern. The global prevalence of chronic kidney disease is approximately 9.1%-13.4%. As chronic kidney disease progresses, the glomerular filtration rate decreases, the quality of life of patients deteriorates, and ultimately end-stage renal disease occurs, imposing a heavy burden on individuals and social economy.

[0003] Renal interstitial fibrosis and cellular crescents are important pathological features in the progression of kidney injury, representing the key links of chronic and acute pathological injuries respectively, and are closely related to the progression and prognosis of chronic kidney disease. Accurately predicting the occurrence of these two pathological changes is of great significance for early intervention and precise treatment of chronic kidney disease. Currently, renal biopsy is mainly relied on for pathological diagnosis in clinical practice to evaluate the degree of renal interstitial fibrosis and cellular crescent injury by microscopy. However, renal biopsy is an invasive operation, which not only has high technical requirements and high costs, but also has low patient acceptance and is difficult to repeat the examination multiple times, which to a certain extent limits its wide application in the management of chronic kidney disease. Therefore, there is an urgent need to establish an efficient and accurate prediction model based on non-invasive biomarkers to provide earlier risk assessment tools and optimize the management strategy of chronic kidney disease. In recent years, the role of immune inflammation in the occurrence and development of chronic kidney disease has received attention, and abnormal lymphocyte function may be involved in the process of kidney injury and fibrosis. In this study, based on traditional indicators, biomarkers related to lymphatic function were further incorporated to construct a more predictive model, so as to improve the early recognition ability of renal interstitial fibrosis and cellular crescent formation, and promote precise intervention and personalized treatment. Summary of the Invention

[0004] In view of the limitations of renal biopsy pathological examination, the present invention provides a precise and efficient prediction model for chronic kidney disease interstitial fibrosis and cellular crescents and its application. The present invention integrates multi-dimensional data such as demographic characteristics, lifestyle, disease history, laboratory test data, renal biopsy pathological information, and lymphocyte function indicators, and uses LASSO regression and Logistic regression to construct a prediction model for chronic kidney disease interstitial fibrosis and cellular crescents, and combines six machine learning methods to verify the model.

[0005] To achieve the above object, the present application adopts the following technical solutions:

[0006] In the first aspect, the present invention provides a prediction model for chronic kidney disease interstitial fibrosis and cellular crescents, and the prediction model includes a prediction model for renal interstitial fibrosis and a prediction model for cellular crescents;

[0007] The renal interstitial fibrosis prediction model includes 5 independent prediction variables, namely estimated glomerular filtration rate, albumin, hematocrit, RBC distribution width SD, and CD4+ / CD8+ T lymphocyte ratio;

[0008] The cellular crescent prediction model includes 6 independent prediction variables, namely age, gender, high-density lipoprotein, estimated glomerular filtration rate, ln(CD3+CD4+ T lymphocyte count), and ln(urinary red blood cell count).

[0009] In the above technical solution, the renal interstitial fibrosis prediction model is:

[0010] Logit(P1) = -0.035 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × albumin (g / L) - 0.039 × hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313;

[0011] Where P1 is the probability value of severe renal interstitial fibrosis;

[0012] The cellular crescent prediction model is:

[0013] Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high-density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / ul) + 0.670 × ln(urinary red blood cell count) ( / ul) + 1.530;

[0014] Where P2 is the probability value of cellular crescent formation, taking the value of 1 when the gender is "male" and 2 when the gender is "female".

[0015] In the above technical solution, the prediction model is visualized using a nomogram.

[0016] In the second aspect, the present invention provides a chronic kidney disease interstitial fibrosis and cellular crescent prediction system, which includes a renal interstitial fibrosis prediction system and a cellular crescent prediction system;

[0017] The renal interstitial fibrosis prediction system includes a variable input module, an analysis module, and an output module;

[0018] The variable input module collects the estimated glomerular filtration rate, albumin, hematocrit, RBC distribution width SD, and CD4+ / CD8+ T lymphocyte ratio of an individual;

[0019] The analysis module is used to calculate the probability value of severe renal interstitial fibrosis. It has a built-in renal interstitial fibrosis prediction model, and the renal interstitial fibrosis prediction model is:

[0020] Logit(P1) = -0.035 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × albumin (g / L) - 0.039 × hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313;

[0021] Where P1 is the probability value of severe renal interstitial fibrosis;

[0022] The output module is used to output the probability value P1 of severe renal interstitial fibrosis obtained by the analysis module;

[0023] The cellular crescent prediction system includes a variable input module, an analysis module, and an output module;

[0024] The variable input module collects the age, gender, high-density lipoprotein, estimated glomerular filtration rate, ln(CD3+CD4+ T lymphocyte count), and ln(urinary red blood cell count) of an individual;

[0025] The analysis module is used to calculate the probability value of cellular crescent formation. It has a built-in cellular crescent prediction model, and the cellular crescent prediction model is:

[0026] Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high-density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / ul) + 0.670 × ln(urinary red blood cell count) ( / ul) + 1.530;

[0027] Where P2 is the probability value of cellular crescent formation. When the gender is "male", the value is 1, and when the gender is "female", the value is 2;

[0028] The output module is used to output the probability value P2 of cellular crescent formation obtained by the analysis module.

[0029] In the above technical solution, the analysis module further includes a nomogram for predicting chronic kidney disease interstitial fibrosis and cellular crescents established based on the information collected by the variable input module.

[0030] In a third aspect, the present invention provides a device for predicting chronic kidney disease interstitial fibrosis and cellular crescents, the prediction device including a renal interstitial fibrosis prediction device and a cellular crescent prediction device;

[0031] The renal interstitial fibrosis prediction device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the following regression equation is run:

[0032] Logit(P1) = -0.035 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × albumin (g / L) - 0.039 × hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313;

[0033] where P1 is the probability value of severe renal interstitial fibrosis;

[0034] The cellular crescent prediction device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the following regression equation is run:

[0035] Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high-density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / ul) + 0.670 × ln(urinary red blood cell count) ( / ul) + 1.530;

[0036] where P2 is the probability value of cellular crescent formation, taking the value of 1 when the gender is "male" and the value of 2 when the gender is "female".

[0037] In a fourth aspect, the present invention provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program. Among them, when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the following regression equation:

[0038] Logit(P1) = -0.035 × estimated glomerular filtration rate (ml / min / 1.73m 2) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC Distribution Width SD (fL) - 0.212 × CD4+ / CD8+ T Lymphocyte Ratio + 0.313;

[0039] Where P1 is the probability value of severe renal interstitial fibrosis;

[0040] Logit(P2) = -0.032 × Age (years) - 0.766 × Gender - 1.047 × High-Density Lipoprotein (mmol / L) - 0.011 × Estimated Glomerular Filtration Rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T Lymphocyte Count) (cells / μl) + 0.670 × ln(Urine Red Blood Cell Count) ( / μl) + 1.530;

[0041] Where P2 is the probability value of cellular crescent formation, taking the value of 1 when the gender is "male" and 2 when the gender is "female".

[0042] In the fifth aspect, the present invention provides an intelligent online prediction tool for chronic renal interstitial fibrosis and cellular crescents. The online prediction tool uses Web technology to achieve risk assessment of renal interstitial fibrosis and cellular crescents. The front end is based on the React or Vue.js framework, and the interface layout is optimized by combining Bootstrap or TailwindCSS; the back end uses Flask or FastAPI to build a computing service, integrates a machine learning model and provides efficient inference capabilities; the front and back ends perform asynchronous data interaction through AJAX or WebSocket. After the front-end user inputs information, the back end calls the renal interstitial fibrosis prediction model function and the cellular crescent prediction model function to calculate in real time and return the results to the front end;

[0043] The renal interstitial fibrosis prediction model function is:

[0044] Logit(P1) = -0.035 × Estimated Glomerular Filtration Rate (ml / min / 1.73m 2 ) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC Distribution Width SD (fL) - 0.212 × CD4+ / CD8+ T Lymphocyte Ratio + 0.313;

[0045] Where P1 is the probability value of severe renal interstitial fibrosis;

[0046] The cellular crescent prediction model function is:

[0047] Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / μl) + 0.670 × ln(urinary red blood cell count) ( / μl) + 1.530;

[0048] Among them, P2 is the probability value of cellular crescent formation. When the gender is "male", the value is 1; when the gender is "female", the value is 2.

[0049] In the sixth aspect, the present invention provides an application of the above prediction model in the preparation of a product for predicting chronic kidney disease interstitial fibrosis and cellular crescent.

[0050] In the study of renal interstitial fibrosis and cellular crescent, by analyzing large sample data, the present invention reveals the important roles of two immune-related biomarkers, the CD4+ / CD8+ T lymphocyte ratio and ln(CD3+CD4+ T lymphocyte count), in renal interstitial fibrosis and cellular crescent formation. This discovery provides new biomarkers for the early diagnosis of renal interstitial fibrosis and cellular crescent, and provides important scientific basis for clinical diagnosis and treatment.

[0051] The present invention relates to a new prediction model for chronic kidney disease interstitial fibrosis and cellular crescent. The construction method of the prediction model includes the following steps:

[0052] S1: The present invention included a total of 2,179 patients who were hospitalized in the nephrology department of a certain hospital in Wuhan from 2021 to 2023 and completed renal biopsy. Demographic characteristics, lifestyle, disease history, laboratory test data, and renal biopsy pathological information of the patients were collected through electronic medical records. Fresh whole blood was collected within 24 hours after the patients were admitted to the hospital. The absolute numbers and proportions of T, B, and NK cells were measured using TruCOUNT reagent tubes, and flow cytometry was used to detect lymph function indicators by FACSCanto flow cytometer (BD, USA).

[0053] S2: After excluding 54 patients under 18 years old and 529 cases of missing data, 1,596 research subjects were included. Among them, 1,203 cases were mild renal interstitial fibrosis (fibrosis ratio ≤ 25%), and 393 cases were severe renal interstitial fibrosis (fibrosis ratio > 25%); 1,209 cases had no cellular crescent, and 387 cases had cellular crescent. Finally, a total of 1,596 research subjects were included for analysis of 90 research variables.

[0054] S3: Comprehensively analyze the 90 variables in S2, and conduct differential analysis between the two groups for the mild interstitial fibrosis group and the mild interstitial fibrosis group, as well as the acellular crescent group and the cellular crescent group, respectively, as follows:

[0055] In step S3, the Shapiro-Wilk normality test method is used to perform normality tests on all continuous variables. Continuous variables with a normal distribution are expressed as mean ± standard deviation, and the t-test is used to compare the differences between groups; variables with a non-normal distribution are expressed as median (M) and interquartile range (Q1, Q3), and the Mann-Whitney U test is used for between-group analysis. Categorical variables are all expressed as the number of cases (percentage), and the χ2 test is used to evaluate the differences between groups. P < 0.05 is considered statistically significant. Based on the above analysis, variables with significant differences between the mild interstitial fibrosis group and the mild interstitial fibrosis group, as well as the acellular crescent group and the cellular crescent group, are respectively screened out as variables to be analyzed.

[0056] S4: Use LASSO regression to analyze the variables to be analyzed screened out in S3:

[0057] (1) 37 potential predictive variables are screened out for renal interstitial fibrosis, namely history of hypertension, estimated glomerular filtration rate, systolic blood pressure, diastolic blood pressure, total cholesterol, procalcitonin, aspartate aminotransferase, calcium, urea, alanine aminotransferase, urine albumin / creatinine ratio, phosphorus, ALT / AST ratio, urine creatinine, total bilirubin, albumin, lactate dehydrogenase, high-sensitivity C-reactive protein, alkaline phosphatase, high-sensitivity cardiac troponin, high-density lipoprotein, mean hemoglobin concentration, RBC distribution width SD, eosinophil count, plateletcrit, hematocrit, RBC distribution width CV, neutrophil count, lymphocyte count, erythrocyte sedimentation rate, neutrophil percentage, total B lymphocyte percentage, CD4+ / CD8+ T lymphocyte ratio, ln(total B lymphocyte count), ln(T / B / NK cell count), CD3+CD8+ T lymphocyte percentage, urine pH;

[0058] (2) 21 potential predictive variables are screened out for renal cellular crescent, namely age, gender, alcohol consumption, education level, estimated glomerular filtration rate, body mass index, total cholesterol, indirect bilirubin, calcium, glucose, triglyceride, alanine aminotransferase, lactate dehydrogenase, γ-glutamyl transpeptidase, high-density lipoprotein, mean PLT volume, hemoglobin, monocyte percentage, ln(CD3+CD4+ T lymphocyte count), CD3+CD4+ T lymphocyte percentage, ln(urine red blood cell count).

[0059] S5: Use Logistic regression to analyze the potential predictive variables selected in S4, screen out the independent influencing variables of renal interstitial fibrosis and cellular crescents, and construct predictive models for renal interstitial fibrosis and cellular crescents respectively. Further evaluate the models through ROC, calibration curves, and DCA curves.

[0060] (1) Renal interstitial fibrosis: Finally, 5 independent predictive variables were screened out, including estimated glomerular filtration rate (OR value: 0.966, 95% CI: 0.961 - 0.970, P < 0.001), albumin (OR value: 1.031, 95% CI: 1.014 - 1.048, P < 0.001), hematocrit (OR value: 0.962, 95% CI: 0.940 - 0.985, P = 0.001), RBC distribution width SD (OR value: 1.037, 95% CI: 1.006 - 1.069, P = 0.020), CD4+ / CD8+ T lymphocyte ratio (OR value: 0.809, 95% CI: 0.705 - 0.923, P = 0.002). The results of the Logistic regression model were visualized using a forest plot.

[0061] Based on the finally selected 5 independent predictive variables, statistical analysis of model parameters was performed to construct a predictive model for renal interstitial fibrosis:

[0062] Logit(P1) = -0.035 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × albumin (g / L) - 0.039 × hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313.

[0063] Where P1 is the probability value of severe renal interstitial fibrosis. The predictive model for renal interstitial fibrosis has been visualized using a nomogram.

[0064] (2) Cellular crescents: A total of 6 independent predictors were finally identified, including age (OR: 0.968, 95% CI: 0.958 - 0.979, P < 0.001), gender (OR: 0.465, 95% CI: 0.352 - 0.611, P < 0.001), high-density lipoprotein (OR: 0.351, 95% CI: 0.240 - 0.506, P < 0.001), estimated glomerular filtration rate (OR: 0.989, 95% CI: 0.985 - 0.993, P < 0.001), ln(CD3+CD4+ T lymphocyte count) (OR: 0.793, 95% CI: 0.674 - 0.934, P = 0.005), and ln(urinary red blood cell count) (OR: 1.954, 95% CI: 1.770 - 2.166, P < 0.001). The results of the Logistic regression model were visualized using a forest plot.

[0065] Based on the 6 finally identified independent predictors, model parameter statistics were performed to construct a prediction model for cellular crescents:

[0066] Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high-density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / μl) + 0.670 × ln(urinary red blood cell count) ( / μl) + 1.530.

[0067] Where P2 is the probability value of cellular crescent formation; it takes the value of 1 when the gender is "male" and 2 when the gender is "female". The prediction model for cellular crescents has been visualized using a nomogram.

[0068] S6: The predictive ability and accuracy of the prediction model were evaluated using ROC curves, calibration curves, and DCA curves.

[0069] S7: Six machine learning methods (random forest, logistic regression tree, decision tree, gradient boosting decision tree, extreme gradient boosting tree, support vector machine) were used to validate the model, and their predictive ability and application value were evaluated through ROC curves.

[0070] The above steps S3 - S7 were all analyzed using R 4.4.2. After the above steps S1 - S7, 5 independent predictors for interstitial fibrosis and 6 independent predictors for cellular crescent formation were finally identified. Further visualization using a nomogram was performed, which is the prediction model for renal interstitial fibrosis and cellular crescents of the present invention.

[0071] Through the foregoing model construction steps, the present invention discloses a new prediction model for interstitial fibrosis and cellular crescents in chronic kidney disease.

[0072] (1) Renal interstitial fibrosis: The five predictive variables are: estimated glomerular filtration rate, albumin, hematocrit, RBC distribution width SD, and CD4+ / CD8+ T lymphocyte ratio; A nomogram is used to visualize the prediction model, specifically including a score scale (0 - 100 points), five interstitial fibrosis predictive variables, total score (0 - 220 points), linear predictor, and probability of severe interstitial fibrosis prediction.

[0073] (2) Renal cellular crescents: The six predictive variables are: age, gender, high - density lipoprotein, estimated glomerular filtration rate, ln(CD3+CD4+ T lymphocyte count), and ln(urinary red blood cell count); A nomogram is used to visualize the prediction model, specifically including a score scale (0 - 100 points), six cellular crescent predictive variables, total score (0 - 220 points), linear predictor, and probability of cellular crescent formation prediction.

[0074] In addition, the present invention provides an intelligent online prediction tool that uses Web technology to achieve risk assessment of renal interstitial fibrosis and cellular crescents. The front - end of the system is based on the React or Vue.js framework, combined with Bootstrap or TailwindCSS to optimize the interface layout to ensure a good user experience. The back - end is built using Flask or FastAPI to provide computing services, integrating machine - learning models and providing efficient inference capabilities. The front - end and back - end perform asynchronous data interaction through AJAX or WebSocket. After the user inputs information, the system calculates in real - time and returns the prediction probability. In addition, this tool can be extended to cloud deployment, suitable for access from different terminal devices, further enhancing the universality and accessibility of the application.

[0075] The prediction model for interstitial fibrosis and cellular crescents in chronic kidney disease of the present invention has important clinical value in aspects such as clinical diagnosis and early screening, clinical decision - making management, personalized treatment formulation, pathological monitoring of renal transplant patients, identification of new predictive indicators, and promotion of multi - scenario applications, as follows:

[0076] (1) Clinical diagnosis and early screening: The present invention can be used as a non - invasive auxiliary tool to help doctors evaluate the risk of renal interstitial fibrosis and cellular crescent formation in patients, improve the early diagnosis rate, and reduce the dependence on renal biopsy, especially suitable for patients who are not suitable for biopsy or have a high risk of biopsy.

[0077] (2) Clinical decision-making management: The prediction model can serve as an auxiliary decision-making tool for clinicians to optimize the diagnosis and treatment plan and conduct risk grading management. When the predicted risk is low, renal biopsy can be postponed, and dynamic follow-up and non-invasive examinations can be used to monitor the condition, avoiding unnecessary invasive procedures and reducing medical risks and economic burdens. If the risk is high, early intervention can be carried out, and renal biopsy can be arranged when necessary to clarify the pathological type and guide individualized treatment, improving the management efficiency of chronic kidney disease.

[0078] (3) Formulation of personalized treatment: The prediction model can provide support for precision medicine, helping clinicians optimize drug regimens, adjust lifestyle intervention measures, and reasonably plan the follow-up frequency, improving the accuracy and effectiveness of disease management and reducing unnecessary hospitalizations and medical expenses.

[0079] (4) Pathological monitoring of kidney transplant patients: The present invention is also applicable to kidney transplant recipients and can be used for long-term monitoring of transplant kidney function, assisting in identifying pathological changes related to chronic transplant nephropathy and immune rejection, and optimizing the follow-up management strategy for transplant patients.

[0080] (5) Identification of new prediction indicators: In the study of renal interstitial fibrosis and cellular crescents, the present invention reveals the important roles of two immune-related biomarkers, the CD4+ / CD8+ T lymphocyte ratio and ln(CD3+CD4+ T lymphocyte count), in the formation of renal interstitial fibrosis and cellular crescents by analyzing large sample data. This discovery provides new biomarkers for the early diagnosis of renal interstitial fibrosis and cellular crescents and important scientific basis for clinical diagnosis and treatment.

[0081] (6) Promotion of multi-scenario applications: The model of the present invention can be widely applied to hospitals, research institutions, health management centers, and online medical platforms to promote the early screening, early diagnosis, and early treatment of chronic kidney disease, ultimately improving the quality of life and life expectancy of patients. 1) Application in clinical departments: Applied in clinical departments such as nephrology, endocrinology, rheumatology and immunology, etc., to assist doctors in formulating precise diagnosis and treatment plans; 2) Promotion in primary healthcare: Improve the management level of chronic kidney disease in primary hospitals, reduce the concentration of patients in tertiary hospitals, and improve the balance of medical resources; 3) Physical examination and health management: Promote in physical examination centers and health management companies to help high-risk groups carry out early intervention and reduce the disease burden; 4) Intelligent remote monitoring: Combined with mobile medical APPs and intelligent wearable devices, enabling patients to assess the risk of kidney disease at any time and enhancing the convenience of chronic disease management.

[0082] The present invention comprehensively analyzes multi-dimensional data such as demographic characteristics, lifestyle, disease history, laboratory test data, renal biopsy pathological information, and lymphocyte function indicators, constructs a precise prediction model for interstitial fibrosis and cellular crescents in chronic kidney disease, and develops an intelligent online prediction tool to achieve efficient and convenient clinical decision-making assistance, and promotes the precise diagnosis and treatment and personalized management of chronic kidney disease.

[0083] Technical principle of the present invention:

[0084] First, based on 1596 inpatients in the nephrology department of a hospital in Wuhan from 2021 to 2023, electronic medical records were used to collect the patients' demographic characteristics, lifestyle, disease history, laboratory test data, and renal biopsy pathological information. Fresh whole blood was collected within 24 hours after all patients were admitted to the hospital, and lymphocyte function indicators were detected by a FACSCanto flow cytometer. According to the fibrosis ratio, chronic kidney disease was divided into a mild renal interstitial fibrosis group (fibrosis ratio ≤ 25%) and a severe renal interstitial fibrosis group (fibrosis ratio > 25%); according to the presence or absence of cellular crescent formation, chronic kidney disease was divided into a non-cellular crescent group and a cellular crescent group. Then, the t-test (for normally distributed variables), Mann-Whitney U test (for non-normally distributed variables), and χ2 test (for categorical variables) were used to screen out variables with significant differences between groups. Through between-group difference analysis, 54 variables with statistical differences in renal interstitial fibrosis and 44 variables with statistical differences in cellular crescents were identified. Subsequently, LASSO regression was used to identify potential predictive indicators that have a key impact on renal interstitial fibrosis and cellular crescents. The present invention uses LASSO regression to introduce a penalty term into the model, shrink the variable coefficients, and screen out 37 potential predictive variables for renal interstitial fibrosis from 54 variables, and 21 potential predictive variables for cellular crescents from 44 variables. Subsequently, the screened variables were further analyzed by Logistic regression, and according to the absolute value of the regression coefficient and the P value, the independent variables that have the most significant impact on the dependent variable were finally determined. The larger the absolute value of the regression coefficient, the more significant the impact of the independent variable on the dependent variable, thereby improving the accuracy and clinical applicability of the model.

[0085] Five independent predictive variables for renal interstitial fibrosis were screened out using Logistic regression, and a prediction model for renal interstitial fibrosis was constructed:

[0086] Logit(P1) = -0.035 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × albumin (g / L) - 0.039 × hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313. Where P1 is the probability value of severe renal interstitial fibrosis.

[0087] Six independent predictors of cellular crescents were screened using logistic regression, and a prediction model for cellular crescents was constructed:

[0088] Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high-density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / μl) + 0.670 × ln(urinary red blood cell count) ( / μl) + 1.530. Here, P2 is the probability value of cellular crescent formation; when the gender is "male", the value is 1, and when the gender is "female", the value is 2.

[0089] Finally, the present invention visualizes the prediction model through a nomogram, providing an intuitive and easy-to-understand way to show the degree of influence of each key variable on the dependent variable. In the nomogram, each predictor variable is assigned a corresponding score according to its influence on the dependent variable, and the total score is calculated by summing the scores of all variables. By establishing the mapping relationship between the total score and the occurrence probability of the dependent variable, the outcome of an individual can be effectively predicted. The nomogram not only simplifies the complex regression equation but also makes the results of the prediction model easier to interpret and apply, thus significantly improving the clinical operability and practical value of the model. With the help of this tool, doctors can quickly and accurately assess the risks of patients in daily clinical practice and provide strong support for personalized treatment plans.

[0090] The present invention provides a prediction model for renal interstitial fibrosis and cellular crescents based on multi-dimensional data, which has significant innovation and clinical application value. Compared with the prior art, the advantages of the present invention are mainly reflected in the following aspects:

[0091] 1. Scientific and rigorous model establishment process

[0092] The present invention first identifies variables with statistical differences through univariate between-group difference analysis, and then adopts LASSO regression, which effectively screens variables and compresses redundant terms by introducing a penalty term, thereby avoiding overfitting and alleviating the problem of multicollinearity, and screening out potential predictive indicators that have a key impact on renal interstitial fibrosis and cellular crescents. Logistic regression analysis is used to finally determine the independent variables that have the most significant influence on the dependent variable. This method ensures the stability and accuracy of the model in a complex clinical environment.

[0093] 2. Multi-dimensional data fusion and accurate prediction

[0094] The present invention not only combines traditional clinical indicators, but also introduces lymphocyte function-related biomarkers, comprehensively considering multi-dimensional information such as renal biopsy pathological information and laboratory test data, thereby constructing a more comprehensive and accurate prediction model. This innovation makes the model more accurate in the early diagnosis and prognosis assessment of renal interstitial fibrosis and cellular crescents.

[0095] 3. Discovery of new prediction indicators

[0096] In the study of renal interstitial fibrosis and cellular crescents, the present invention first discovered that the ratio of CD4+ / CD8+ T lymphocytes serves as an independent prediction indicator for severe renal interstitial fibrosis, and ln(CD3+CD4+ T lymphocyte count) serves as an independent prediction indicator for the formation of renal interstitial fibrosis, providing new ideas and directions for subsequent research.

[0097] 4. Reliable prediction ability

[0098] The present invention comprehensively verified the model using six mainstream machine learning methods (including random forest, logistic regression tree, decision tree, gradient boosting decision tree, extreme gradient boosting tree, and support vector machine). The verification of different algorithms ensured the accuracy, generalization, and credibility of the model, verifying the good prediction ability of the model.

[0099] 5. Non-invasive and efficient

[0100] The present invention breaks through the traditional method that relies on renal biopsy, adopts non-invasive diagnostic means, and constructs a non-invasive and accurate prediction model by collecting patients' demographic characteristics, lifestyle, laboratory data, and lymphocyte function indicators. Compared with traditional invasive detection methods, it greatly reduces the pain and medical expenses of patients, and at the same time avoids the limitations brought by problems such as the high technical difficulty of renal biopsy and low patient acceptance.

[0101] 6. Simple and operable nomogram visualization tool

[0102] The present invention visualizes the prediction model through a nomogram, simplifies the complex regression equation, enabling doctors to intuitively understand the impact of each variable on the formation of renal interstitial fibrosis and cellular crescents. The nomogram not only enhances the operability of the model, but also provides a convenient prediction of the probability of individual kidney pathological damage through the mapping relationship between the total score and the probability of the dependent variable.

[0103] 7. Promote precise intervention and personalized treatment

[0104] The present invention helps doctors accurately evaluate the risks of patients at an early stage and formulate personalized intervention and treatment plans based on the prediction results. This is of great significance for optimizing the management strategies of chronic kidney disease, improving the effectiveness of early intervention, and reducing the incidence of end-stage renal disease. Especially for kidney transplant patients, this model helps in the long-term monitoring of transplanted kidney function and pathological changes related to immune rejection.

[0105] 8. Broad clinical application prospects

[0106] The prediction model of the present invention is applicable to various medical institutions, such as hospitals, research institutions, health management centers, and online medical platforms. It can be widely applied in multiple clinical departments such as nephrology, endocrinology, rheumatology and immunology, etc., improving the management level of chronic kidney disease in primary hospitals, reducing the medical treatment costs of patients, and promoting the rational allocation of medical resources.

[0107] In summary, the present invention greatly improves the prediction ability of renal interstitial fibrosis and cellular crescents by comprehensively integrating multi-dimensional data, introducing innovative model analysis methods, and combining with easy-to-operate visualization tools. This technology not only provides strong support for the early identification, precise intervention, and personalized treatment of chronic kidney disease, but also has significant clinical application value. Compared with the existing technology, the present invention has higher accuracy, operability, and application prospects, and has important scientific and practical significance. Brief description of the drawings

[0108] Figure 1 is the flow chart for constructing the prediction model of renal interstitial fibrosis and cellular crescents of the present invention.

[0109] Figure 2 is the flow cytometry gating strategy for the functions of T / B / NK lymphocytes of the present invention; wherein A: total white blood cell gating; B: B cell gating; C: T cell gating; D: helper / inducer and suppressor / cytotoxic T lymphocyte gating; E: NK cell gating.

[0110] Figure 3 are representative pathological pictures of mild and severe renal interstitial fibrosis of the present invention; wherein A: mild interstitial fibrosis; B: severe interstitial fibrosis.

[0111] Figure 4 is the representative pathological picture of cellular crescent formation of the present invention; wherein A: no cellular crescent formation; B: cellular crescent formation.

[0112] Figure 5 is the screening process of LASSO regression analysis for renal interstitial fibrosis of the present invention; wherein A: LASSO coefficient path diagram; B: LASSO cross-validation curve.

[0113] Figure 6The screening process of LASSO regression analysis for cellular crescents in the present invention; wherein A: LASSO coefficient path diagram; B: LASSO cross-validation curve.

[0114] Figure 7 The forest plot of the Logistic regression for the renal interstitial fibrosis prediction model of the present invention.

[0115] Figure 8 The forest plot of the Logistic regression for the cellular crescent prediction model of the present invention.

[0116] Figure 9 The nomogram of the renal interstitial fibrosis prediction model of the present invention.

[0117] Figure 10 The nomogram of the cellular crescent prediction model of the present invention.

[0118] Figure 11 The evaluation of the renal interstitial fibrosis prediction model of the present invention; A: ROC curve; B: calibration curve; C: DCA curve.

[0119] Figure 12 The evaluation of the cellular crescent prediction model of the present invention; A: ROC curve; B: calibration curve; C: DCA curve.

[0120] Figure 13 The ROC curve diagram of the 6 machine learning methods for validating the renal interstitial fibrosis prediction model of the present invention.

[0121] Figure 14 The ROC curve diagram of the 6 machine learning methods for validating the cellular crescent prediction model of the present invention. Detailed implementation manners

[0122] To better illustrate the purpose, technical solution and advantages of the present invention, the present invention will be further described below in conjunction with specific embodiments. The present invention can be implemented in many different forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the present invention to those skilled in the art. The present invention will only be defined by the claims.

[0123] Example 1 Construction of prediction models for chronic kidney disease interstitial fibrosis and cellular crescents based on multi-dimensional data of inpatients in the nephrology department of a hospital in Wuhan

[0124] S1. In the embodiments of the present invention, 2179 patients who were hospitalized in the Nephrology Department of a certain hospital in Wuhan from 2021 to 2023 and received renal biopsy were included. Demographic characteristics, lifestyle, medical history, laboratory test results, and renal biopsy pathological information of the patients were collected through the electronic medical record system. Fresh whole blood was collected within 24 hours after the patients were admitted to the hospital, and the lymphocyte function indexes of the patients were detected using a FACSCanto flow cytometer (BD, USA). The flow cytometry gating strategy for T / B / NK lymphocyte function is shown in Figure 2 .

[0125] S2. In this embodiment, after excluding the inclusion criteria, a total of 1596 subjects and 90 research variables were included in the analysis. Among them, 1203 cases had mild renal interstitial fibrosis (fibrosis ratio ≤ 25%), and 393 cases had severe renal interstitial fibrosis (fibrosis ratio > 25%); 1209 cases had no cellular crescents, and 387 cases had cellular crescents. Representative pathological images of mild and severe renal interstitial fibrosis are shown in Figure 3 , and the representative pathological image of cellular crescent formation is shown in Figure 4 .

[0126] Population inclusion criteria: (1) Diagnosed with chronic kidney disease; (2) Completed renal biopsy during hospitalization.

[0127] Population exclusion criteria: (1) Excluded cases under 18 years of age; (2) Excluded cases with missing values.

[0128] Definition of chronic kidney disease: Estimated glomerular filtration rate < 60 mL / min / 1.73m 2 , lasting for ≥ 3 months; or ACR (urinary albumin / creatinine ratio) ≥ 30 μg / mg.

[0129] Data collection:

[0130] Electronic medical record data: age, gender, marital status, family history of chronic kidney disease, history of hypertension, history of diabetes, history of coronary heart disease, history of stroke, smoking, alcohol consumption, education level, ln(serum creatinine), estimated glomerular filtration rate, systolic blood pressure, diastolic blood pressure, body mass index, urinary total microprotein, total cholesterol, albumin / globulin ratio, indirect bilirubin, low-density lipoprotein, 24-hour urinary total microprotein, procalcitonin, uric acid, aspartate aminotransferase, calcium, urea, glucose, triglyceride, alanine aminotransferase, urinary albumin / creatinine ratio, phosphorus, ALT / AST ratio, urinary creatinine, total bilirubin, urinary microalbumin, albumin, total protein, lactate dehydrogenase, high-sensitivity C-reactive protein, gamma-glutamyl transpeptidase, 24-hour urinary microalbumin, alkaline phosphatase, high-sensitivity cardiac troponin, high-density lipoprotein, direct bilirubin, N-terminal pro-brain natriuretic peptide, globulin, monocyte count, mean corpuscular hemoglobin concentration, RBC distribution width SD, white blood cell count, platelet count, eosinophil count, plateletcrit, basophil count, hematocrit, mean platelet volume, mean RBC volume, RBC distribution width SD, RBC distribution width CV, red blood cell count, mean corpuscular hemoglobin, hemoglobin, platelet distribution width, neutrophil count, monocyte percentage, lymphocyte percentage, eosinophil percentage, large platelet ratio, lymphocyte count, basophil percentage, erythrocyte sedimentation rate, neutrophil percentage, ln(urinary white blood cell count), urine pH, ln(urinary red blood cell count); Detection of whole blood lymphatic function indicators: total T lymphocyte percentage, total B lymphocyte percentage, T / B / NK cell percentage, NK cell percentage, CD4+ / CD8+ T lymphocyte ratio, ln(CD3+CD4+ T lymphocyte count), ln(total B lymphocyte count), CD3+CD4+ T lymphocyte percentage, ln(T / B / NK cell count), CD3+CD8+ T lymphocyte percentage, ln(CD3+CD8+ T lymphocyte count), ln(total T lymphocyte count), ln(NK cell count).

[0131] S3. Use R4.4.2 to perform an inter-group difference analysis on 90 variables in the mild interstitial fibrosis group and the mild interstitial fibrosis group, as well as the acellular crescent group and the cellular crescent group. In this embodiment, the Shapiro-Wilk normality test method is used to perform a normality test on all continuous variables. If P is less than 0.05, it is considered that the data does not conform to the normal distribution. The t-test and the Mann-Whitney U test are used to complete the continuous variables with normal distribution and the continuous variables with non-normal distribution respectively; the chi-square test is used to complete the difference analysis of the categorical variables between groups; if P is less than 0.05, the difference is considered to be statistically significant. The inter-group difference analysis between the mild and severe renal interstitial fibrosis groups is shown in Table 1; the inter-group difference analysis between the acellular and cellular crescent groups is shown in Table 2.

[0132] Table 1. Variables Included in the Study of Renal Interstitial Fibrosis and Their Differential Analysis (N = 1596)

[0133]

[0134]

[0135]

[0136]

[0137] Table 2. Variables Included in the Study of Renal Cellular Crescent and Their Differential Analysis (N = 1596)

[0138]

[0139]

[0140]

[0141] S4. Further analyze the variables with inter-group differences screened in S3 using LASSO regression. In this example, the glmnet package in R 4.4.2 software was used to complete the LASSO regression analysis. By performing LASSO regression on 54 variables with statistical differences in renal interstitial fibrosis and 44 variables in cellular crescents, the variable coefficients were shrunk, and 37 potential predictive variables ( Figure 5 A and B) were screened out for renal interstitial fibrosis, and 21 potential predictive variables ( Figure 6 A and B) were screened out for renal cellular crescents.

[0142] S5. Analyze the potential predictive variables screened in S4 using Logistic regression to screen out the independent influencing variables for renal interstitial fibrosis and cellular crescents. Further, perform statistical analysis on the model parameters of the predictive variables to construct a prediction model.

[0143] In this example, 5 independent predictive variables for renal interstitial fibrosis were finally screened out through the Logistic regression model, including estimated glomerular filtration rate, albumin, hematocrit, RBC distribution width SD, and CD4+ / CD8+ T lymphocyte ratio, and a forest plot was used to show the influence degree of the predictive variables on renal interstitial fibrosis ( Figure 7 ). In this example, 6 independent predictive variables for renal cellular crescents were finally screened out through the Logistic regression model, including age, gender, high-density lipoprotein, estimated glomerular filtration rate, ln(CD3+CD4+ T lymphocyte count), and ln(urinary red blood cell count), and a forest plot was used to show the influence degree of the predictive variables on renal cellular crescents ( Figure 8 ). Further, a nomogram was used for renal interstitial fibrosis (Figure 9 ) and cellular crescents ( Figure 10 ) for visualization. In this example, R 4.4.2 software and its related R packages were used for data analysis and visualization. Among them, the stats package was used to complete the Logistic regression analysis, forestplot was used to draw the forest plot, and the rms package was used to draw the nomogram.

[0144] S6. Use the ROC curve, calibration curve, and DCA curve to evaluate the predictive ability and accuracy of the prediction model. In this example, the areas under the ROC curves of renal interstitial fibrosis and cellular crescents were 0.848 ( Figure 11 A) and 0.806 ( Figure 12 A), indicating that the prediction model has good predictive ability. The calibration curves of renal interstitial fibrosis ( Figure 11 B) and cellular crescents ( Figure 12 B) follow the trend of y = x, and the mean absolute error values are 0.028 and 0.039 respectively, indicating that the risk prediction model can be largely consistent with the actual observed values. The DCA curve shows that when the probabilities of severe interstitial fibrosis and cellular crescent formation are between 0 and 1, the prediction model shows good net benefits. In this example, the pROC package in R 4.4.2 software was used to draw the ROC curve. The Bootstrap method was used to internally validate the prediction model, and the calibration curve was drawn and evaluated after 1000 repeated samplings. The drawing of the calibration curve was completed by the rms package; the drawing of the DCA curve was completed by the rmda package.

[0145] S7. Six machine learning methods (random forest, logistic regression tree, decision tree, gradient boosting decision tree, extreme gradient boosting tree, support vector machine) were used to validate the renal interstitial fibrosis and cellular crescent models. In this example, the dataset was divided into a training set and a test set in a ratio of 7:3, and the prediction model was repeatedly validated using five-fold cross-validation combined with six machine learning methods. The AUC areas of the six machine learning methods showed that the validation results of each model indicated that the prediction models for renal interstitial fibrosis ( Figure 13 ) and cellular crescents ( Figure 14 ) had good sensitivity and specificity. The validation results of multiple machine learning models showed that the prediction model in the present invention had excellent predictive ability.

[0146] The present invention provides an important tool for the prediction models of chronic kidney disease interstitial fibrosis and cellular crescents in clinical diagnosis and early screening, clinical decision-making management, personalized treatment formulation, pathological monitoring of renal transplant patients, identification of new predictive indicators, and promotion of multi-scenario applications.

[0147] Obviously, the above embodiments are merely examples for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to enumerate all implementation manners here. And the obvious changes or variations derived therefrom still fall within the protection scope of the present invention.

Claims

1. A prediction model for interstitial fibrosis and cellular crescent in chronic kidney disease, characterized in that: The prediction model includes a renal interstitial fibrosis prediction model and a cellular crescent prediction model; The renal interstitial fibrosis prediction model includes 5 independent prediction variables, namely estimated glomerular filtration rate, albumin, hematocrit, RBC distribution width SD, and CD4+ / CD8+ T lymphocyte ratio; The cellular crescent prediction model includes 6 independent prediction variables, namely age, gender, high-density lipoprotein, estimated glomerular filtration rate, ln(CD3+CD4+ T lymphocyte count), and ln(urinary red blood cell count).

2. The prediction model according to claim 1, characterized in that: The renal interstitial fibrosis prediction model is: Logit(P1) = -0.035 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313; where P1 is the probability value of severe renal interstitial fibrosis; The cellular crescent prediction model is: Logit(P2) = -0.032 × age (years) - 0.766 × gender - 1.047 × high-density lipoprotein (mmol / L) - 0.011 × estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / μl) + 0.670 × ln(urinary red blood cell count) ( / μl) + 1.530; where P2 is the probability value of cellular crescent formation, taking the value of 1 when the gender is "male" and 2 when the gender is "female".

3. The prediction model according to claim 1, wherein: The prediction model is visualized using a nomogram.

4. A prediction system for interstitial fibrosis and cellular crescents in chronic kidney disease, characterized in that: The prediction system includes a renal interstitial fibrosis prediction system and a cellular crescent prediction system; The renal interstitial fibrosis prediction system includes a variable input module, an analysis module, and an output module; The variable input module collects the estimated glomerular filtration rate, albumin, hematocrit, RBC distribution width SD, and CD4+ / CD8+ T lymphocyte ratio of an individual; The analysis module is used to calculate the probability value of severe renal interstitial fibrosis. It has a built-in renal interstitial fibrosis prediction model, and the renal interstitial fibrosis prediction model is: Logit(P1) = -0.035 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313; where P1 is the probability value of severe renal interstitial fibrosis; The output module is used to output the probability value P1 of severe renal interstitial fibrosis obtained by the analysis module; The cellular crescent prediction system includes a variable input module, an analysis module, and an output module; The variable input module collects the age, gender, high-density lipoprotein, estimated glomerular filtration rate, ln(CD3+CD4+ T lymphocyte count), and ln(urinary red blood cell count) of an individual; The analysis module is used to calculate the probability value of cellular crescent formation. It has a built-in cellular crescent prediction model, and the cellular crescent prediction model is: Logit(P2) = -0.032 × Age (years) - 0.766 × Gender - 1.047 × High - density lipoprotein (mmol / L) - 0.011 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T - lymphocyte count) (cells / μl) + 0.670 × ln(Urine red blood cell count) ( / μl) + 1.530; where P2 is the probability value of cellular crescent formation, taking the value of 1 when the gender is "male" and 2 when the gender is "female"; The output module is used to output the probability value P2 of cellular crescent formation obtained by the analysis module.

5. The prediction system according to claim 4, wherein: The analysis module also includes a nomogram for predicting chronic kidney disease interstitial fibrosis and cellular crescent based on the information collected by the variable input module.

6. A prediction device for interstitial fibrosis and cellular crescent in chronic kidney disease, characterized in that: The prediction device includes a renal interstitial fibrosis prediction device and a cellular crescent prediction device; The renal interstitial fibrosis prediction device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it runs the following regression equation: Logit(P1) = -0.035 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313; where P1 is the probability value of severe renal interstitial fibrosis; The cellular crescent prediction device includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it runs the following regression equation: Logit(P2) = -0.032 × Age (years) - 0.766 × Gender - 1.047 × High-density lipoprotein (mmol / L) - 0.011 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T lymphocyte count) (cells / μl) + 0.670 × ln(Urine red blood cell count) ( / μl) + 1.530; Where P2 is the probability value of cellular crescent formation, which takes the value of 1 when the gender is "male" and 2 when the gender is "female".

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein, when the computer program runs, it controls the device where the computer-readable storage medium is located to execute the regression equation described below: Logit(P1) = -0.035 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313; Where P1 is the probability value of severe renal interstitial fibrosis; Logit(P2) = -0.032 × Age (years) - 0.766 × Gender - 1.047 × High - density lipoprotein (mmol / L) - 0.011 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T - lymphocyte count) (cells / μl) + 0.670 × ln(Urine red blood cell count) ( / μl) + 1.530; Where P2 is the probability value of cellular crescent formation, which takes the value of 1 when the gender is "male" and 2 when the gender is "female".

8. An intelligent online prediction tool for interstitial fibrosis and cellular crescent in chronic kidney disease, characterized in that: The online prediction tool uses Web technology to achieve the risk assessment of renal interstitial fibrosis and cellular crescent. The front end is based on the React or Vue.js framework, and combines Bootstrap or TailwindCSS to optimize the interface layout; the back end uses Flask or FastAPI to build a computing service, integrates machine learning models and provides efficient inference capabilities; the front and back ends perform asynchronous data interaction through AJAX or WebSocket. After the front-end user inputs information, the back end calls the renal interstitial fibrosis prediction model function and the cellular crescent prediction model function to calculate in real time and return the results to the front end; The renal interstitial fibrosis prediction model function is: Logit(P1) = -0.035 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) + 0.030 × Albumin (g / L) - 0.039 × Hematocrit (%) + 0.036 × RBC distribution width SD (fL) - 0.212 × CD4+ / CD8+ T lymphocyte ratio + 0.313; Where P1 is the probability value of severe renal interstitial fibrosis; The cellular crescent prediction model function is: Logit(P2) = -0.032 × Age (years) - 0.766 × Gender - 1.047 × High - density lipoprotein (mmol / L) - 0.011 × Estimated glomerular filtration rate (ml / min / 1.73m 2 ) - 0.233 × ln(CD3+CD4+ T - lymphocyte count) (cells / μl) + 0.670 × ln(Urine red blood cell count) ( / μl) + 1.530; Where P2 is the probability value of cellular crescent formation, which takes the value of 1 when the gender is "male" and 2 when the gender is "female".

9. Use of the prediction model according to any one of claims 1-3 in the preparation of a product for predicting chronic renal interstitial fibrosis and cellular crescent.

Images