CN112837819B - Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation - Google Patents

Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation Download PDF

Info

Publication number
CN112837819B
CN112837819B CN202110077761.7A CN202110077761A CN112837819B CN 112837819 B CN112837819 B CN 112837819B CN 202110077761 A CN202110077761 A CN 202110077761A CN 112837819 B CN112837819 B CN 112837819B
Authority
CN
China
Prior art keywords
risk factors
kidney injury
acute kidney
coronary artery
bypass grafting
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110077761.7A
Other languages
Chinese (zh)
Other versions
CN112837819A (en
Inventor
侯剑峰
林宏远
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fuwai Hospital of CAMS and PUMC
Original Assignee
Fuwai Hospital of CAMS and PUMC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Fuwai Hospital of CAMS and PUMC filed Critical Fuwai Hospital of CAMS and PUMC
Priority to CN202110077761.7A priority Critical patent/CN112837819B/en
Publication of CN112837819A publication Critical patent/CN112837819A/en
Application granted granted Critical
Publication of CN112837819B publication Critical patent/CN112837819B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • GPHYSICS
    • G16INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16HHEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

The invention discloses a method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting, which is characterized in that a risk assessment model is established by collecting and sorting a large number of domestic patient hospital data and combining direct single factor and indirect single factor regression screening and a multi-factor regression optimization analysis method, and the accuracy and reliability of the assessment model are effectively improved by increasing the sample size and introducing new risk factors, especially by increasing alanine aminotransferase and introducing brain natriuretic peptide, so that an effective assessment prediction method is provided for acute kidney injury risk assessment of coronary artery bypass grafting of heart failure patients in China, and the method is of great significance to health development of medical level in China.

Description

Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation
Technical Field
The invention belongs to the field of biomedicine, and particularly relates to a method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting.
Background
Coronary heart disease is the most common cause of heart failure, and Coronary Artery Bypass Graft (CABG) surgery is one of the important methods for surgical treatment of coronary heart disease with cardiac insufficiency. However, due to the high requirements of the surgical technique and the complexity of perioperative management, serious complications of patients in perioperative period are high, and among a plurality of complications, acute Kidney Injury (AKI) has high incidence rate, and postoperative multi-organ dysfunction, most commonly heart insufficiency and AKI, has close relation with perioperative death and postoperative life quality reduction. Currently, a number of predictive scoring systems have been established for renal insufficiency after cardiac surgery, with the cleveland ARF score, the Mehta score, and the SRI score being more common. However, these models were designed primarily for all patient populations for cardiac surgery, and most were based on clinical data 10 years ago, data were collected primarily from western national populations. There may be a bias in the assessment of patients receiving simple CABG today, especially cardiac insufficiency. Therefore, a risk assessment model considering Chinese crowd specificity and accuracy is established, postoperative adverse events can be accurately predicted, high-risk patients are identified by layering risk factors, perioperative risk factors are controlled, and then the purposes of reducing complications and improving medical quality are achieved, so that the method has important clinical significance.
Disclosure of Invention
The invention aims to provide a method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting, which can more accurately evaluate the risk of acute kidney injury after coronary artery bypass grafting of heart failure patients, is a model more suitable for risk prediction of Chinese people, and is suitable for popularization and application in clinical practice.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
a method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting operation is characterized in that the model firstly adopts direct single factor regression analysis on risk factors of acute kidney injury after coronary artery bypass grafting operation, then obtains combined risk factors through the indirect single factor regression analysis method, finally further screens the combined risk factors through multi-factor regression analysis to obtain final risk factors, and determines a line-list prediction model.
Further, the building of the model comprises the following steps:
(1) Collecting clinical data of patients with statistics of preoperative cardiac insufficiency and patients receiving coronary artery bypass grafting operation to establish a disease database;
(2) Carrying out correlation analysis on 27 possible risk factors respectively, firstly carrying out direct single factor logistic regression analysis, and determining the direct risk factors which are ranked at the front relative to the acute kidney injury;
(3) Taking the risk factors which are ranked ahead in the step (2) as dependent variables respectively, performing indirect single-factor logistic regression analysis on the rest variables, and determining the indirect risk factors ranked ahead again;
(4) Combining the risk factors ranked at the front in the step (2) and the step (3) to obtain a combined risk factor;
(5) Carrying out multi-factor logistic regression analysis on the combined risk factors, deleting the co-linear variable, and obtaining final risk factors related to the postoperative acute kidney injury;
(6) Assigning a value to each final risk factor based on its regression coefficient, and establishing a line graph prediction model equation for the incidence of acute kidney injury: p (t) =λ 0 (t)exp(β 1 x 12 x 2 +…+β k x k );
(7) The model is externally validated by a validation set.
Further, all the 27 possible risk factor variables in the step (2) are classified variables, and are expressed by frequency (percentage).
Further, the direct risk factors of step (2) are 3: creatinine, history of myocardial infarction, and peri-operative transfusion.
Further, the indirect risk factors of step (3) are 10: sex, diabetes, elevated alanine aminotransferase, heart surgery, brain natriuretic peptide, peripheral arterial lesions, left ventricular ejection fraction, hypertension, chronic obstructive pulmonary disease, and extracorporeal circulation surgery.
Further, the combined risk factors of step (4) are 13: creatinine, history of myocardial infarction, perioperative transfusion, gender, diabetes, elevated alanine aminotransferase, cardiac surgery, brain natriuretic peptide, peripheral arterial lesions, left ventricular ejection fraction, hypertension, chronic obstructive pulmonary disease, and extracorporeal circulation surgery, the final risk factors of step (5) are 9: sex, elevated alanine aminotransferase, brain natriuretic peptide, blood creatinine, left ventricular ejection fraction, history of myocardial infarction, hypertension, extracorporeal circulation surgery, and perioperative blood transfusion.
Further, the direct single-factor Logistic regression analysis in the step (2) and the indirect single-factor Logistic regression analysis in the step (3) are performed, and the significance of the risk factors is ranked according to the absolute value of the accuracy.
Further, the multi-factor Logistic regression analysis in the step (5) adopts an Enter method, and a line list chart is built based on a Logistic regression equation.
Further, the external verification of step (7) includes discrimination assessment using AUC values, model calibration assessment using calibration curves, and net benefit rate assessment using decision curve analysis.
Further, all statistical analyses were done in R language (version 3.5).
Compared with the existing other models, the method for establishing the acute kidney injury prediction model after coronary artery bypass grafting is characterized in that Chinese patients are used as research objects in the model establishing process, the collected disease population cardinality is investigated, database samples are abundant, and the investigated risk factors are more, so that the model is more suitable for Chinese populations and is higher in prediction accuracy compared with the existing other prediction models, and is suitable for evaluating postoperative acute kidney injury risks in clinical practice.
The prediction model establishment process disclosed by the invention combines direct single factor regression with indirect single factor regression analysis, and further optimizes the screening result through multi-factor regression analysis, and screens out 9 final risk factors.
Particularly, by the step-by-step screening method, the independent risk factors alanine aminotransferase and brain natriuretic peptide are introduced into the prediction model, so that the model shows better differentiation degree, calibration degree and net benefit rate, and has important significance for improving the prediction accuracy of the model; the improvement of the prediction accuracy also indirectly proves that the rise of alanine aminotransferase and the occurrence of acute kidney injury in the perioperative period of brain natriuretic peptide have a direct risk relation, so that more reference and evaluation basis are provided for the perioperative risk evaluation of coronary artery bypass grafting of heart failure patients; meanwhile, two risk factors of alanine aminotransferase elevation and brain natriuretic peptide are introduced and proposed, and the method has important guiding and reference significance for further researching the risk reasons and mechanisms of acute kidney injury in the perioperative period of coronary artery bypass grafting of heart failure patients.
Drawings
FIG. 1 is a schematic diagram of an alignment chart prediction model of the present invention;
FIG. 2 is a schematic diagram showing the comparison of calibration curves of various models of the building block of the present invention;
FIG. 3 is a graph showing the comparison of the working curves (ROC) of model subjects of the present invention;
FIG. 4 is a graphical representation of the comparative curves of the working curves (ROC) of the model subjects of the validation set of the present invention;
FIG. 5 is a schematic diagram showing analysis and comparison of decision curves of various models of the building block of the present invention;
FIG. 6 is a schematic diagram of analysis of decision curves of various models of the verification set of the present invention;
FIG. 7 is a flow chart of a method for establishing a predictive model of acute kidney injury after coronary artery bypass grafting according to the present invention.
Detailed Description
The invention discloses a method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting, the flow diagram of which is shown in fig. 7, and the invention is further described and illustrated with reference to the accompanying drawings.
Example 1
The method for collecting and sorting the patients with heart failure receiving coronary artery bypass grafting operation in 3659 domestic with complete clinical data as modeling study objects in 2010 to 2019 comprises the following steps: as study subjects 27 risk factors were used, gender, hyperlipidemia, brain natriuretic peptide, thyroid function, hemoglobin, alanine aminotransferase, hypertension, body mass index, history of myocardial infarction, diabetes, stent implantation of cardiac vessels, elevated creatinine, cardiac surgery, history of smoking, peripheral arterial lesions, cerebrovascular events, pre-operative critical conditions, CCS grade 4, preoperative atrial fibrillation or flutter, NYHA cardiac function grade III or IV, left ventricular ejection fraction (LVEF < 35%), concomitant valve surgery, concomitant aortic surgery, non-selective surgery, chronic obstructive pulmonary disease, extracorporeal circulation surgery and perioperative blood transfusion.
The definition of acute kidney injury is based on the following three criteria, one of which is satisfied, that can be diagnosed as acute kidney injury: (1) The increase in serum creatinine (SCr) within 48 hours is greater than or equal to 0.3mg/dL; (2) an increase in SCr of 1.5 times or more the basal value within 7 days; (3) urine volume of less than 0.5ml/kg/h for 6 consecutive hours.
The hyperlipidemia judgment standard is that when the following fasting plasma examination index is more than or equal to 1 item, dyslipidemia can be diagnosed, total Cholesterol (TC) is more than or equal to 6.2mmol/L, low density lipoprotein cholesterol (LDL-C) is more than or equal to 4.1mmol/L, triglyceride (TG) is more than or equal to 2.3mmol/L, and high density lipoprotein cholesterol (HDL-C) is less than 1.0mmol/L; the brain natriuretic peptide classification criteria are: less than 50 years old, brain natriuretic peptide > 450pg/ml, between 50 and 75 years old, brain natriuretic peptide > 900pg/ml, greater than 75 years old, brain natriuretic peptide greater than 1800pg/ml; thyroid function, whether there is a history of thyroid abnormalities; hemoglobin, with < 90g/L as demarcation point; alanine aminotransferase, whether there is an increase in alanine aminotransferase; hypertension, whether there is a systolic pressure > 140mmHg or a diastolic pressure > 90mmHg; whether there is a history of myocardial infarction; whether diabetes has a history of sugar sickness; the stent implantation of the cardiac blood vessel is carried out in the prior art, and whether the stent implantation operation of the cardiac blood vessel is carried out or not; blood creatinine, preoperative blood creatinine > 176umol/L; heart surgery, whether there is heart surgery with open pericardium; a history of smoking, whether there is a history of smoking; peripheral arterial lesions, whether peripheral arterial lesions exist in the past; cerebrovascular events, with or without coma or central nervous system abnormalities for more than 24 hours, for more than 72 hours; a pre-operative critical state, whether any of ventricular tachycardia or ventricular fibrillation or sudden death from rescue; CCS grade 4, CCS angina classification grade 4; preoperative atrial fibrillation or atrial augmentation, and preoperative atrial fibrillation or atrial augmentation is not present in two weeks; left ventricular ejection fraction (LVEF < 35%); combining valve operations, whether there are any valve operations combined; merging aortic surgery, whether there are any merging aortic surgery; non-chronology surgery, whether or not there is a non-chronology surgery; chronic obstructive pulmonary disease, whether there is an excessive chronic obstructive pulmonary disease; an extracorporeal circulation operation, whether the extracorporeal circulation operation is performed; a perioperative blood transfusion, whether or not there is a perioperative blood transfusion; but also sex and body mass index. And (5) arranging the patient information and establishing a patient information database.
Example 2
Statistical analysis: all variables are classified variables, the variables are expressed by frequency (percentage), and in single factor analysis, the significance of risk factors is ordered according to the absolute value of the correct rate, and the P value is required to be smaller than 0.1. The multi-factor Logistic regression analysis adopts an 'Enter' method which is an optional analysis method in a regression analysis method column, and a linear chart prediction model is established based on a Logistic regression equation.
Dividing 3659 patients into two groups according to age, sex and body quality index indexes, wherein one group serves as a building module, the other group serves as a verification group, the number of the modeling group is 2365, the number of the verification group is 1294, and the distribution conditions of the age, sex and body quality index indexes of the building module and the verification group are basically consistent.
In a building module (n=2365), 27 risk factors are integrated in the research by combining the previous research and clinical experience, direct single factor logistic regression analysis is firstly carried out, the correlation between 27 risk factors and acute kidney injury is directly analyzed, the correlation significance of the risk factors is ordered according to the absolute value of the accuracy, the P value requirement is less than 0.1, and 3 direct risk factors with the highest order are obtained relative to the acute kidney injury: and after the completion of the blood creatinine, myocardial infarction medical history and perioperative blood transfusion, performing indirect single factor logistic regression analysis, namely respectively taking the blood creatinine, myocardial infarction medical history and perioperative blood transfusion as dependent variables, performing indirect single factor regression analysis on the rest 24 variables, wherein the significance of the risk factors is still ordered according to the absolute value of the accuracy, the P value requirement is smaller than 0.1, and selecting the corresponding 4 indirect risk factors with the earlier ordering after the completion of each variable regression analysis, wherein four risk factors of diabetes, alanine aminotransferase elevation, peripheral arterial lesions and brain natriuretic peptide are obtained from blood creatine analysis results, four risk factors of LVEF < 35%, chronic obstructive pulmonary disease, external circulation surgery and peripheral arterial lesions are obtained from the myocardial infarction medical history analysis results, and four risk factors of heart surgery, gender, hypertension and brain natriuretic peptide are obtained from perioperative blood transfusion analysis results. A total of 10 indirect risk factors were screened by the above analysis: sex, diabetes, elevated alanine aminotransferase, heart surgery, brain natriuretic peptide, peripheral arterial lesions, LVEF < 35%, hypertension, chronic obstructive pulmonary disease and extracorporeal circulation surgery. The 10 indirect risk factors mentioned above together with the serum creatinine, history of myocardial infarction and 3 direct risk factors for perioperative transfusion give a total of 13 combined risk factors: creatinine, history of myocardial infarction, perioperative transfusion, gender, diabetes, elevated alanine aminotransferase, cardiac surgery, brain natriuretic peptide, peripheral arterial lesions, LVEF < 35%, hypertension, chronic obstructive pulmonary disease, and extracorporeal circulation surgery, and for these 13 combined risk factors, a multifactor logistic regression analysis was performed to delete the co-linear variables, determining 9 final risk factors associated with postoperative acute kidney injury: sex, elevated alanine aminotransferase, brain natriuretic peptide, creatinine, LVEF < 35%, history of myocardial infarction, hypertension, extracorporeal circulation surgery and peri-operative blood transfusion. The 9 final risk factors correspond to weights and coefficients as shown in table 1.
TABLE 1 risk factors and weights
Independent risk factors selected on the basis of multi-factor analysis are established, assignment is carried out on the independent risk factors based on regression coefficients of each independent risk factor, see table 1, a linear chart prediction model (figure 1) is established by using a logistic regression method, and the occurrence rate of acute kidney injury is: p (t) =λ 0 (t)exp(β 1 x 12 x 2 +…+β k x k ) Wherein P (t) is the incidence of acute kidney injury, lambda 0 (t) is a function of computing AKI, x 1 、x 2 ...x k Beta as covariates i Is x i Is used for predicting the incidence of acute kidney injury after surgery.
The model was evaluated for calibration, discrimination and net benefit rate by plotting a calibration curve (calibration curve), ROC subject work curve and decision curve analysis (decision curve analysis, DCA). And compared to the currently clinically used cleveland ARF score, mehta score, and SRI score.
The model calibration was evaluated using a calibration curve, the discrimination was evaluated using an AUC value, the net benefit rate was evaluated using a DCA curve, all statistical analyses were done using R language (version 3.5), and P values less than 0.05 were considered statistically significant.
Comparing the calibration curves of the models of the building module with the calibration curves of the other three models (figure 2), the calibration curves of the present linear graph prediction model and the present linear graph prediction model show that the present linear graph prediction model has better calibration degree (figure 2) than other models, and particularly has high calibration degree in the range that the predicted AKI occurrence rate is lower than 30 percent.
The model subjects working curve (ROC) for the model set was compared to the schematic (fig. 3), in which the current line graph prediction model AUC was 0.807, the cliff score AUC was 0.638, the mehta score AUC was 0.597, the sri score AUC was 0.600, the validation set was compared to the model subjects working curve (ROC) (fig. 4), in which the current line graph prediction model AUC was 0.815, the cliff score AUC was 0.612, the mehta score AUC was 0.620, the sri score AUC was 0.572, the area under the model subject working curve (AUC) was higher for the current line graph than for the other 3 common models, the AUC was the area enclosed by the axis under the ROC curve, the closer to 1.0, the higher the detection method was true, the higher the discrimination was, and the more accurate the prediction result was.
The model decision curve analysis and comparison schematic diagram (figure 5) of the building module and the model decision curve analysis schematic diagram (figure 6) of the verification group show that the model predictive model of the linear array shows higher net benefit rate in the building module and the verification group than the other three models.
A total of 2365 patients are established, wherein 205 patients with AKI after operation (8.67%) are included in the linear chart prediction model, and 9 independent risk factors such as gender, alanine aminotransferase, brain natriuretic peptide, pre-operation blood creatinine increase, LVEF < 35%, past myocardial infarction, hypertension, extracorporeal circulation operation and perioperative blood transfusion are included in the linear chart prediction model. The verification group comprises 1294 patients, wherein 119 AKI (9.19%) of the patients are generated after operation, the linear chart prediction model shows better differentiation degree, calibration degree and net benefit rate than other 3 models, and the linear chart prediction model has higher prediction accuracy on acute kidney injury after coronary artery bypass grafting operation of patients with cardiac insufficiency.
Example 3
The influence experiment of the independent risk factors alanine aminotransferase and brain natriuretic peptide on the prediction model of the line chart verifies that the object is a modeling group crowd, when the prediction model of the line chart is incorporated into gender, the pre-operation blood creatinine is increased, the LVEF is less than 35 percent, 7 independent risk factors such as myocardial infarction, hypertension, extracorporeal circulation surgery, perioperative blood transfusion and the like are adopted, and the area under the working curve (AUC) of the subject is 0.738; when the model incorporates sex, pre-operative creatinine is increased, LVEF is less than 35%, 8 independent risk factors such as past myocardial infarction, hypertension, extracorporeal circulation surgery, perioperative blood transfusion, alanine aminotransferase and the like, and the area under the working curve (AUC) of a subject is 0.762; when the model is incorporated into gender, the pre-operation creatinine is increased, the LVEF is less than 35%, and 8 independent risk factors such as myocardial infarction, hypertension, extracorporeal circulation operation, perioperative blood transfusion and brain natriuretic peptide are present, and the area under the working curve (AUC) of a subject is 0.751; when the model incorporates sex, pre-operative creatinine is increased, LVEF is less than 35%, 9 independent risk factors such as past myocardial infarction, hypertension, extracorporeal circulation surgery, perioperative blood transfusion, alanine aminotransferase and brain natriuretic peptide are included, and the area under the working curve (AUC) of a subject is 0.807; therefore, in the model building process, the inclusion of the independent risk factors alanine aminotransferase and brain natriuretic peptide has important significance for the prediction model of the present line list graph to show better differentiation, calibration degree and net benefit rate.
The foregoing examples are provided to facilitate understanding of the method and core ideas of the present application, and are not to be construed as limiting the present application where variations are apparent to those of ordinary skill in the art in light of the concepts of the present application.

Claims (9)

1. A method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting operation is characterized in that a prediction equation of the prediction model is that:P(t)=λ 0 (t)exp(β 1 x 12 x 2 +…+β k x k ) The method comprises the steps of carrying out a first treatment on the surface of the The P (t) is the incidence rate of acute kidney injury, lambda 0 (t) is a function of computing AKI, x 1 、x 2 ...x k Beta as covariates i Is x i Regression coefficients of (a);
a method for establishing a prediction model of acute kidney injury after coronary artery bypass grafting operation includes the steps that direct single factor regression analysis is firstly adopted for risk factors of acute kidney injury after coronary artery bypass grafting operation, then combined risk factors are obtained through an indirect single factor regression analysis method, finally the combined risk factors are further screened through multi-factor regression analysis to obtain final risk factors, and a line-drawing prediction model is determined;
the establishment of the prediction model comprises the following steps:
(1) Collecting clinical data of patients with statistics of preoperative cardiac insufficiency and patients receiving coronary artery bypass grafting operation to establish a disease database;
(2) Carrying out correlation analysis on 27 possible risk factors respectively, firstly carrying out direct single factor logistic regression analysis, and determining the direct risk factors which are ranked at the front relative to the acute kidney injury;
(3) Taking the risk factors which are ranked ahead in the step (2) as dependent variables respectively, performing indirect single-factor logistic regression analysis on the rest variables, and determining the indirect risk factors ranked ahead again;
(4) Combining the direct risk factors and the indirect risk factors in the step (2) and the step (3) to obtain combined risk factors;
(5) Carrying out multi-factor logistic regression analysis on the combined risk factors, deleting the co-linear variable, and obtaining final risk factors related to the postoperative acute kidney injury;
(6) Assigning a value to each final risk factor based on the regression coefficient of each final risk factor, and establishing a line list prediction model equation;
(7) Externally verifying the model through a verification group;
the line drawing prediction model in the step (6) comprises 9 independent risk factors including sex, alanine aminotransferase, brain natriuretic peptide, pre-operation blood creatinine increase and LVEF < 35%, past myocardial infarction, hypertension, extracorporeal circulation operation and perioperative blood transfusion.
2. The method for constructing a predictive model of acute kidney injury after coronary artery bypass grafting according to claim 1, wherein the 27 possible risk factor variables in step (2) are all classified variables and are expressed by frequency numbers.
3. A method of modeling acute kidney injury after coronary artery bypass grafting according to claim 1, wherein the direct risk factors of step (2) are 3: creatinine, history of myocardial infarction, and peri-operative transfusion.
4. A method of modeling acute kidney injury after coronary artery bypass grafting according to claim 1, wherein the indirect risk factors of step (3) are 10: sex, diabetes, elevated alanine aminotransferase, heart surgery, brain natriuretic peptide, peripheral arterial lesions, left ventricular ejection fraction, hypertension, chronic obstructive pulmonary disease, and extracorporeal circulation surgery.
5. A method of modeling acute kidney injury after coronary artery bypass grafting according to claim 1, wherein the combined risk factors of step (4) are 13: creatinine, history of myocardial infarction, perioperative transfusion, gender, diabetes, elevated alanine aminotransferase, cardiac surgery, brain natriuretic peptide, peripheral arterial lesions, left ventricular ejection fraction, hypertension, chronic obstructive pulmonary disease, and extracorporeal circulation surgery, the final risk factors of step (5) are 9: sex, elevated alanine aminotransferase, brain natriuretic peptide, blood creatinine, left ventricular ejection fraction, history of myocardial infarction, hypertension, extracorporeal circulation surgery, and perioperative blood transfusion.
6. The method for establishing a predictive model of acute kidney injury after coronary artery bypass grafting according to claim 1, wherein the direct single factor Logistic regression analysis of step (2) and the indirect single factor Logistic regression analysis of step (3) are characterized in that the significance of risk factors is ordered according to the absolute value of the accuracy.
7. The method for building a predictive model of acute kidney injury after coronary artery bypass grafting according to claim 1, wherein the multi-factor Logistic regression analysis in step (5) uses an "Enter" method and builds a line graph based on Logistic regression equations.
8. A method of modeling acute kidney injury prediction after coronary artery bypass grafting according to claim 1, wherein the external verification of step (7) includes discrimination assessment using AUC values, model calibration assessment using calibration curves, and net benefit rate assessment using decision curve analysis.
9. A method of modeling acute kidney injury after coronary bypass grafting according to claim 1 wherein all statistical analyses are performed in R language.
CN202110077761.7A 2021-01-20 2021-01-20 Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation Active CN112837819B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110077761.7A CN112837819B (en) 2021-01-20 2021-01-20 Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110077761.7A CN112837819B (en) 2021-01-20 2021-01-20 Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation

Publications (2)

Publication Number Publication Date
CN112837819A CN112837819A (en) 2021-05-25
CN112837819B true CN112837819B (en) 2024-04-16

Family

ID=75929197

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110077761.7A Active CN112837819B (en) 2021-01-20 2021-01-20 Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation

Country Status (1)

Country Link
CN (1) CN112837819B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112820397B (en) * 2021-01-20 2024-04-05 中国医学科学院阜外医院 Method for establishing peri-operative risk prediction model of coronary artery bypass grafting
CN113591582B (en) * 2021-07-01 2024-05-31 山东科技大学 Character recognition device and method based on resting state functional magnetic resonance data
CN113409945A (en) * 2021-07-02 2021-09-17 上海市第一人民医院 Model capable of predicting acute kidney injury caused by sepsis
CN113921136A (en) * 2021-09-13 2022-01-11 吾征智能技术(北京)有限公司 System for fusing multi-source data to intelligently evaluate and predict chronic disease risk
CN113782211A (en) * 2021-10-27 2021-12-10 首都医科大学附属北京地坛医院 Establishment method and prediction method of risk early warning model for chronic kidney disease of hepatitis B cirrhosis patients
CN114974595A (en) * 2022-05-13 2022-08-30 江苏省人民医院(南京医科大学第一附属医院) Crohn's disease patient mucosa healing prediction model and method
CN115910365B (en) * 2022-11-16 2024-02-13 华中科技大学同济医学院附属同济医院 Construction method and prediction system of Eos-CRSwNP preoperative refractory evaluation model
CN117894466B (en) * 2024-03-15 2024-05-17 四川大学华西医院 Continuous acute nephritis risk prediction method and system based on recurrent neural network

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101689220A (en) * 2007-04-05 2010-03-31 奥利安实验室有限公司 The system and method that be used for the treatment of, diagnosis and prospective medicine illness takes place
CN110827992A (en) * 2019-11-15 2020-02-21 袁洪 Preoperative prediction method for acute renal injury after hypertension operation
CN111613334A (en) * 2020-06-01 2020-09-01 广东省心血管病研究所 Method for establishing acute kidney injury probability prediction model after aortic arch replacement
CN112017783A (en) * 2020-09-14 2020-12-01 华中科技大学同济医学院附属协和医院 Prediction model for pulmonary infection after heart operation and construction method thereof

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101689220A (en) * 2007-04-05 2010-03-31 奥利安实验室有限公司 The system and method that be used for the treatment of, diagnosis and prospective medicine illness takes place
CN110827992A (en) * 2019-11-15 2020-02-21 袁洪 Preoperative prediction method for acute renal injury after hypertension operation
CN111613334A (en) * 2020-06-01 2020-09-01 广东省心血管病研究所 Method for establishing acute kidney injury probability prediction model after aortic arch replacement
CN112017783A (en) * 2020-09-14 2020-12-01 华中科技大学同济医学院附属协和医院 Prediction model for pulmonary infection after heart operation and construction method thereof

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
体外循环对冠状动脉旁路移植术后患者早期急性肾损伤的影响;席雷;王正清;刘志刚;梁少杰;滕磊;;第三军医大学学报(第12期);全文 *
冠状动脉旁路移植术后急性肾损伤的发生情况与危险因素分析;杨立猛等;中国血液净化(第11期);第738-741页 *

Also Published As

Publication number Publication date
CN112837819A (en) 2021-05-25

Similar Documents

Publication Publication Date Title
CN112837819B (en) Method for establishing acute kidney injury prediction model after coronary artery bypass grafting operation
CN111312401B (en) After-physical-examination chronic disease prognosis system based on multi-label learning
CN115019919B (en) Deep learning-based method for establishing function prediction model after early recovery of stroke
CN113327679A (en) Pulmonary embolism clinical risk and prognosis scoring method and system
CN114898873A (en) Method and system for predicting cardiovascular disease risk of diabetes mellitus pre-stage patient
CN112820397B (en) Method for establishing peri-operative risk prediction model of coronary artery bypass grafting
CN114220540A (en) Construction method and application of diabetic nephropathy risk prediction model
CN111883258B (en) Method for constructing OHSS indexing parting prediction model
CN113470816A (en) Machine learning-based diabetic nephropathy prediction method, system and prediction device
CN114242245A (en) Machine learning method, system and device for predicting diabetic nephropathy occurrence risk based on electronic medical record data
CN114974585A (en) Construction method of early risk prediction and evaluation model of metabolic syndrome in gestational period
CN114373544A (en) Method, system and device for predicting membranous nephropathy based on machine learning
CN113744832B (en) Intelligent decision-making and quality control system for continuous kidney substitution therapy
CN117493970A (en) Deep learning-based prediction method for adverse cardiovascular events after PCI (peripheral component interconnect) operation of coronary heart disease patient
Farzi et al. Predicting serious diabetic complications using hidden pattern detection
CN117174304A (en) Survival prognosis prediction model for patients with chronic obstructive pulmonary disease combined with pulmonary arterial hypertension
CN116564512A (en) Artificial intelligence prognosis evaluation system for heart failure and renal insufficiency patient
CN114400094A (en) Construction method and application of myocardial infarction risk prediction model after percutaneous coronary intervention treatment for multi-disease co-morbid old patients
Avaliani et al. Prognostic value of plasma myeloperoxidase level’s and echocardiographic determinants in chronic heart failure patients
Zhou et al. A novel laboratory-based model to predict the presence of obstructive coronary artery disease comparison to coronary artery disease consortium 1/2 score, duke clinical score and diamond-forrester score in china
CN115547489A (en) Method for predicting death risk of maintenance hemodialysis patient
CN115691788A (en) Dual attention coupling network diabetes classification system based on heterogeneous data
Cong-Cong et al. Development and validation of a nomogram for postoperative severe acute kidney injury in acute type A aortic dissection
Arias et al. Predictive Model and Risk Score for In-Hospital Mortality in Patients with All-Cause Cardiogenic Shock Santiago Shock Score (S3)
Ceh-Varela et al. Machine learning analysis of factors contributing to Diabetes Development

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
TA01 Transfer of patent application right

Effective date of registration: 20210903

Address after: 100037 Beijing City, Xicheng District No. 167 North

Applicant after: FUWAI HOSPITAL, CHINESE ACADEMY OF MEDICAL SCIENCES

Address before: 6-3, 417 Hengshan street, Lushunkou District, Dalian, Liaoning, 116041

Applicant before: Yin Lijun

TA01 Transfer of patent application right
GR01 Patent grant
GR01 Patent grant