Method and system for establishing prediction model of acute myocardial infarction prognosis

Abstract

The invention discloses an acute myocardial infarction prognosis prediction model establishment method and system, and relates to the technical field of risk prediction.The method comprises the steps that patient chip data is collected and screened, and a first myocardial infarction key gene is obtained; verifying the first myocardial infarction key gene to obtain a second myocardial infarction key gene; analyzing the relationship between the expression level of the key gene and the prognosis of the patient based on the second myocardial infarction key gene; and constructing an acute myocardial infarction prognosis prediction model by using the relationship between the expression level of the key gene and the prognosis of the patient, and performing performance evaluation. By combining immune-related LncRNA, a second myocardial infarction key gene is screened out, an acute myocardial infarction prognosis prediction model is constructed, the occurrence rate of MACE events of a patient is effectively predicted, the accuracy of myocardial infarction prognosis prediction is improved, high-risk patients are screened, and more effective prevention and treatment are provided.

Description

Technical Field

[0001] The invention relates to the technical field of risk prediction, and in particular to a method and system for establishing a prediction model for the prognosis of acute myocardial infarction. Background Art

[0002] Acute myocardial infarction (AMI) is a cardiovascular disease that seriously threatens human life and health. This disease not only directly endangers the patient's life, but also has a long-term impact on the quality of life of survivors. The pathological mechanism of acute myocardial infarction (AMI) mainly involves the sudden interruption of coronary blood flow, which leads to myocardial ischemia and hypoxia, and then triggers a series of complex pathophysiological changes, such as myocardial cell necrosis, inflammatory response, oxidative stress and ventricular remodeling. Although revascularization and drug therapy can improve the symptoms of acute myocardial infarction (AMI) patients to a certain extent and reduce the mortality rate in recent years, the mortality rate, stroke risk or re-hospitalization rate within the first 6 months after discharge is still as high as 25%. Therefore, the current clinical focus is on how to reduce the mortality of patients with acute myocardial infarction. The mortality rate in the acute phase is high, but this is far from the treatment endpoint of this disease. How to effectively prevent reinfarction after PCI for acute myocardial infarction, minimize myocardial damage, reduce adverse cardiovascular events, inhibit ventricular remodeling and improve prognosis is still a clinical problem that needs to be solved urgently; because these long-term complications not only seriously affect the quality of life and survival of patients, but also bring a huge burden to the medical system, therefore, finding markers that can predict the occurrence of adverse cardiovascular events in patients with acute myocardial infarction after PCI and establishing a suitable prognostic model, and then timely giving high-risk patients early intervention, reducing medical costs, and improving patients' quality of life, is of great clinical value in reducing patients' readmission rate and postoperative complication rate.

[0003] However, the current common solutions have many shortcomings, including: although existing studies have found that some clinical or laboratory results are related to the prognosis of patients after PCI for acute myocardial infarction, there is no unanimously recognized predictive model for predicting the prognosis of patients after PCI for acute ST-segment elevation myocardial infarction, which seriously restricts medical workers from early identification and timely intervention of high-risk patients, increases medical costs, and prevents the efficient use of medical resources. Summary of the invention

[0004] In view of the fact that there is no unanimously recognized prediction model for predicting the prognosis of acute myocardial infarction after PCI in the prior art, which seriously restricts medical workers from early identification and timely intervention of high-risk patients and the inability to efficiently utilize medical resources, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is how to provide a marker that can predict the occurrence of adverse cardiovascular events in patients with acute myocardial infarction after PCI and construct a suitable prognostic model, so as to provide timely early intervention for high-risk patients and reduce the patient's readmission rate and the incidence of postoperative complications.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a method for establishing a prediction model for the prognosis of acute myocardial infarction, which includes collecting and screening patient chip data to obtain a first myocardial infarction key gene; verifying the first myocardial infarction key gene to obtain a second myocardial infarction key gene; analyzing the relationship between the key gene expression level and the patient's prognosis based on the second myocardial infarction key gene; using the relationship between the key gene expression level and the patient's prognosis to construct an acute myocardial infarction prognosis prediction model and perform performance evaluation.

[0008] As a preferred scheme of the method for establishing a prediction model for the prognosis of acute myocardial infarction described in the present invention, the first myocardial infarction key gene includes the following steps: downloading an acute myocardial infarction patient chip data set based on the same platform from a database; performing background correction and quantile normalization on the acute myocardial infarction patient chip data set, and drawing a graph for the corrected data; performing differential expression analysis on the corrected data to obtain differentially expressed genes; performing GO analysis and KEGG pathway enrichment analysis on the differentially expressed genes, and then predicting the protein interaction PPI network to mine the key genes in the differentially expressed genes; screening out M key genes before myocardial infarction to obtain the first myocardial infarction key gene.

[0009] As a preferred embodiment of the method for establishing a prediction model for the prognosis of acute myocardial infarction described in the present invention, the second myocardial infarction key gene includes the following steps: selecting A patients with acute myocardial infarction and dividing them into a myocardial infarction group; selecting A patients with normal coronary angiography and dividing them into a healthy control group; using RT-qPCR technology to detect the expression level of the first myocardial infarction key gene in the myocardial infarction group and the healthy control group, that is, detecting the level of plasma LncRNA, screening out the top N key genes consistent with the actual clinical situation, and obtaining the second myocardial infarction key gene.

[0010] As a preferred embodiment of the method for establishing a prediction model for the prognosis of acute myocardial infarction described in the present invention, the analysis of the relationship between the expression level of key genes and the prognosis of patients includes the following steps: using RT-qPCR technology to detect B patients with acute myocardial infarction to obtain the expression level of the second key gene of myocardial infarction; following up the B patients with acute myocardial infarction for C years to observe the patients' adverse cardiovascular events, i.e., MACE events; combining the expression level of the second key gene of myocardial infarction with the patients' adverse cardiovascular events to obtain the relationship between the expression level of the key gene and the prognosis of the patients.

[0011] As a preferred embodiment of the method for establishing a prediction model for the prognosis of acute myocardial infarction described in the present invention, the MACE events include angina attacks, recurrent acute myocardial infarction, target vessel revascularization, worsening of cardiac function and the incidence of all-cause mortality.

[0012] As a preferred embodiment of the method for establishing a prediction model for the prognosis of acute myocardial infarction described in the present invention, the construction of the prediction model for the prognosis of acute myocardial infarction includes the following steps: based on the Harrell guideline for multivariate analysis sample size, the sample size is collected according to the rule that the number of event occurrences EPV> the first threshold; based on the sample size, a univariate logistic regression analysis is performed to obtain the P value of the variable and make a judgment. If the P value of the variable is < the second threshold, the corresponding variable is determined to be associated with the MACE event and recorded to obtain the first prediction variable; if the P value of the variable is ≥ the second threshold, the corresponding variable is determined to be unrelated to the MACE event; the first prediction variable is analyzed by the stepwise regression method to obtain the Akaike information criterion AIC of the first prediction variable and make a judgment. If the Akaike information criterion AIC of the first prediction variable reaches the minimum, the corresponding variable is divided into the second prediction variable; the first prediction variable is analyzed by the optimal subset regression method to obtain the adjusted R of the first prediction variable. 2 , Bayesian Information Criterion BIC and Malos CP value and make judgments. If the adjusted R 2 reaches the maximum, the corresponding variable is divided into the maximum adjustment R 2 predictor variable; if the Bayesian information criterion BIC of the first predictor variable reaches the minimum, the corresponding variable is divided into the predictor variable with the minimum Bayesian information criterion BIC; if the Malos CP value of the first predictor variable reaches the minimum, the corresponding variable is divided into the predictor variable with the minimum Malos CP value; based on the maximum adjusted R 2The prediction variables of the minimum Bayesian information criterion BIC and the prediction variables of the minimum Malos CP value are taken as the union to obtain the third prediction variable; the first prediction variable is analyzed by LASSO regression method to obtain the lambda value of the mean square error and make a judgment. If the lambda value of the mean square error reaches the minimum, it is determined that the number of independent variables is the least, and the corresponding non-zero regression coefficient variable is divided into the fourth prediction variable; based on the intersection of the second prediction variable, the third prediction variable and the fourth prediction variable, the model inclusion variable is obtained; based on the model inclusion variable, a multifactor Logistic regression analysis is performed to obtain the regression coefficient of the model inclusion variable and assign values ​​to the risk factors to construct an acute myocardial infarction prognosis prediction model; according to the acute myocardial infarction prognosis prediction model, a Nomogram model is constructed and the model is displayed.

[0013] As a preferred embodiment of the method for establishing the prediction model for the prognosis of acute myocardial infarction of the present invention, the calculation formula of the prediction model for the prognosis of acute myocardial infarction is as follows:

[0014] T=λ0+λ1×q1+λ2×q2+...+λ n ×q n

[0015] Wherein, T is the risk prediction score of MACE events after PCI for acute myocardial infarction; λ0 is the regression coefficient of the intercept of the prognostic prediction model for acute myocardial infarction; λ1, λ2 and λ n are the regression coefficients of the first, second and nth variables included in the model; q1, q2 and q n They are the 1st, 2nd and nth variables included in the model respectively.

[0016] In the second aspect, the present invention further solves the safety problems existing in the prediction of the prognosis of acute myocardial infarction. The embodiment provides a system for establishing a prediction model for the prognosis of acute myocardial infarction, which includes: a gene screening module, which is used to collect and screen the chip data set of patients with acute myocardial infarction, mine the key genes in the differentially expressed genes, and obtain the first myocardial infarction key gene; a gene verification module, which is used to verify the expression level of the first myocardial infarction key gene using the myocardial infarction group and the healthy control group, and obtain the second myocardial infarction key gene; a relationship analysis module, which is used to detect patients with acute myocardial infarction, obtain the expression level of the second myocardial infarction key gene and follow up, and observe the MACE events of patients with acute myocardial infarction; a model construction module, which is used to use univariate Logistic regression analysis, stepwise regression method, optimal subset regression method, LASSO regression method to screen variables, obtain model variables, and construct an acute myocardial infarction prognosis prediction model based on multivariate Logistic regression analysis, use the Nomogram model nomogram to display the model, and evaluate the performance of the model based on model performance evaluation, model verification and model methodology quality evaluation.

[0017] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the method for establishing a predictive model for the prognosis of acute myocardial infarction as described in the first aspect of the present invention.

[0018] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the method for establishing a predictive model for the prognosis of acute myocardial infarction as described in the first aspect of the present invention.

[0019] The beneficial effects of the present invention are as follows: the present invention screens out the second key myocardial infarction gene and constructs an acute myocardial infarction prognosis prediction model by combining immune-related LncRNA, which can effectively predict the incidence of MACE events in patients, improve the accuracy of myocardial infarction prognosis prediction, screen high-risk patients and provide more effective prevention and treatment, improve the diagnosis and treatment effects of acute myocardial infarction patients, and reduce the waste of medical resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0021] Figure 1 This is a research flow chart of the acute myocardial infarction prognosis prediction model in Example 1.

[0022] Figure 2 For Example 2, with maximum adjustment R 2 Incorporate time variables into process diagrams.

[0023] Figure 3 For Example 2, with maximum adjustment R 2 Schematic diagram of the variables present in the model.

[0024] Figure 4 This is a diagram of the variable inclusion process in Example 2 when the minimum Bayesian Information Criterion BIC is achieved.

[0025] Figure 5 Schematic diagram of the variables present in the model in Example 2 when the model has the minimum Bayesian Information Criterion BIC.

[0026] Figure 6 This is a diagram of the variable incorporation process in Example 2 when the minimum Malos CP value is achieved.

[0027] Figure 7 Schematic diagram of the variables present in the model when the model has the minimum Malos CP value in Example 2.

[0028] Figure 8 This is the coefficient path diagram of the risk variables in the LASSO regression analysis in Example 2.

[0029] Fig. 9 It is a cross-validation curve diagram in the LASSO regression analysis in Example 2.

[0030] Fig.10 This is a nomogram model for the risk of MACE events in patients with acute myocardial infarction after PCI in Example 2.

[0031] Fig.11 This is the ROC curve diagram of the acute myocardial infarction prognosis prediction model in Example 2.

[0032] Fig.12 This is a calibration curve diagram of the prediction model for the risk of ISR in patients after stent implantation in Example 2.

[0033] Fig.13 This is a decision curve diagram of the prediction model for the risk of ISR in patients after stent implantation in Example 2.

[0034] Fig.14 This is the ROC curve diagram of the random split verification in Example 2.

[0035] Fig.15This is a calibration curve diagram for random split verification in Example 2.

[0036] Fig.16 This is a calibration curve diagram for cross validation in Example 2.

[0037] Fig.17 This is the calibration curve diagram for Bootstrap sampling verification in Example 2. DETAILED DESCRIPTION

[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0039] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0040] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0041] Example 1

[0042] Reference Figure 1 , which is the first embodiment of the present invention, provides a method for establishing a prediction model for the prognosis of acute myocardial infarction, comprising the following steps:

[0043] S1: Collect and screen the patient's chip data to obtain the first key gene for myocardial infarction.

[0044] Specifically, Figure 1The figure shows the research process of the acute myocardial infarction prognosis prediction model. The chip data set of acute myocardial infarction patients was collected and screened to obtain the first myocardial infarction key gene. The RT-qPCR technology was used to detect the expression level of the first myocardial infarction key gene in the myocardial infarction group and the healthy control group to obtain the second myocardial infarction key gene. The acute myocardial infarction patients were followed up to observe the patients' adverse cardiovascular events. Then, the variables were screened by using univariate regression and multivariate regression to screen the variables, and then the variables were screened and combined by stepwise regression, optimal subsets and LASSO regression to obtain the variables included in the model. The acute myocardial infarction prognosis prediction model and Nomogram model nomogram were constructed using the included model variables, and then the model was displayed. Subsequently, the performance of the model was evaluated from the perspectives of discrimination, calibration and clinical utility. Secondly, the models were verified one by one using multiple internal validation methods such as random split validation, cross validation and Bootstrap sampling validation. Finally, PROBAST was used to evaluate the methodological quality and the bias risk of the model to achieve the establishment of a prediction model for the prognosis of acute myocardial infarction.

[0045] Preferably, the first myocardial infarction key gene includes the following steps: downloading 4 acute myocardial infarction patient chip data sets GSE29111, GSE48060, GSE66360 and GSE97320 based on the same platform GPL570 from the GEO database.

[0046] The Affy package of R / Bioconductor was used to perform background correction and quantile normalization on the chip dataset of patients with acute myocardial infarction, and the corrected data were plotted, including box plots and density curves.

[0047] The corrected data were analyzed for differential expression using the linear model microarray analysis limma package of R / Bioconductor to obtain differentially expressed genes.

[0048] The DAVID database was used to perform GO analysis and KEGG pathway enrichment analysis on the differentially expressed genes, and the STRING tool was used to predict the protein interaction PPI network to further explore the key genes in the differentially expressed genes, among which GO analysis was a functional analysis.

[0049] The cytoHubba plug-in in the Cytoscape software was used to screen out the top 10 key genes for myocardial infarction and obtain the first key gene for myocardial infarction.

[0050] Specifically, by integrating multiple public data sets and screening out key genes related to myocardial infarction, the genetic characteristics related to the disease can be captured more comprehensively, and the accuracy and reliability of subsequent prediction models can be improved. Compared with a single data set or traditional experimental methods, the present invention can reduce bias and enhance the accuracy of the results.

[0051] S2: Verify the first key gene for myocardial infarction and obtain the second key gene for myocardial infarction.

[0052] Preferably, the second myocardial infarction key gene includes the following steps: selecting 10 patients with acute myocardial infarction and dividing them into a myocardial infarction group.

[0053] Ten patients with normal coronary angiography were selected and divided into healthy control group.

[0054] RT-qPCR technology was used to detect the expression level of the first key gene for myocardial infarction in the myocardial infarction group and the healthy control group, that is, the level of plasma LncRNA was detected, and the top three key genes NEAT1, KCNQ1OT1 and XIST that were consistent with the actual clinical situation were screened out, and the second key gene for myocardial infarction was obtained.

[0055] Specifically, by using RT-qPCR technology to verify the screened key genes in clinical samples, we can further ensure the relevance of the selected genes in actual patients, make up for the limitations that may exist in relying solely on bioinformatics analysis, and enhance the clinical relevance and reliability of the research results.

[0056] S3: Analysis of the key genes for second myocardial infarction and the relationship between key gene expression levels and patient prognosis.

[0057] Preferably, analyzing the relationship between the expression level of key genes and patient prognosis includes the following steps: using RT-qPCR technology to detect 200 patients with acute myocardial infarction to obtain the expression level of the second myocardial infarction key gene.

[0058] 200 patients with acute myocardial infarction were followed up for one year to observe their adverse cardiovascular events, namely MACE events, with the occurrence of MACE events in patients as the observation endpoint.

[0059] The expression levels of the key genes for the second myocardial infarction were combined with the patient's adverse cardiovascular events to obtain the relationship between the expression levels of the key genes and the patient's prognosis.

[0060] Specifically, MACE events include angina attack, recurrent acute myocardial infarction, target vessel revascularization, worsening cardiac function, and all-cause mortality.

[0061] Preferably, through prospective follow-up studies, gene expression levels are associated with MACE events, which not only verifies the predictive value of gene markers, but also provides direct clinical data support for the construction of prediction models, which can better evaluate the predictive ability of gene markers compared with studies based only on retrospective data.

[0062] S4: Use the relationship between key gene expression levels and patient prognosis to construct an acute myocardial infarction prognosis prediction model and evaluate its performance.

[0063] Preferably, constructing a prognosis prediction model for acute myocardial infarction includes the following steps: based on the Harrell guidelines for sample size for multivariate analysis, sample size is collected according to the rule that event occurrence EPV>10, preferably 20 event occurrence EPVs, that is, the number of outcome variables ISR is at least 10 to 20 times the number of predictor variables included in the model, to ensure good accuracy in evaluating the regression coefficient.

[0064] Based on the sample size, a univariate logistic regression analysis was performed to obtain the P value of the variable and make a judgment. If the P value of the variable was less than the second threshold, the corresponding variable was determined to be associated with the MACE event and recorded to obtain the first predictor variable.

[0065] If the P value of the variable is ≥ the second threshold, it is determined that the corresponding variable is not related to the MACE event.

[0066] The first predictor variable is analyzed by using the stepwise regression method to obtain the Akaike information criterion AIC of the first predictor variable and make a judgment. If the Akaike information criterion AIC of the first predictor variable reaches the minimum, the corresponding variable in the model is divided into the second predictor variable.

[0067] The first predictor variable is analyzed using the best subset regression method to obtain the adjusted R of the first predictor variable. 2 , Bayesian Information Criterion BIC and Malos CP value and make judgments. If the adjusted R 2 reaches the maximum, the corresponding variable in the model is divided into the maximum adjusted R 2 predictor variables.

[0068] If the Bayesian Information Criterion (BIC) of the first predictor variable reaches the minimum, the corresponding variable in the model is classified as the predictor variable with the minimum Bayesian Information Criterion (BIC).

[0069] If the Malos CP value of the first predictor variable reaches the minimum, the corresponding variable in the model is classified as the predictor variable with the minimum Malos CP value.

[0070] Based on the maximum adjustment R 2The third predictor is obtained by taking the union of the predictor variable with the minimum Bayesian information criterion BIC and the predictor variable with the minimum Malos CP value.

[0071] The predictor obtained by univariate logistic regression analysis was used as the independent variable, and whether the outcome variable ISR occurred was used as the dependent variable. The LASSO regression method was used to analyze the first predictor variable, and the lambda value of the mean square error was obtained and judged. If the lambda value of the mean square error reached the minimum, it was judged that the model had excellent performance and the number of independent variables was the least, and the corresponding non-zero regression coefficient variable was divided into the fourth predictor variable.

[0072] The variables included in the model are obtained by taking the intersection of the second predictor variable, the third predictor variable and the fourth predictor variable.

[0073] Multivariate Logistic regression analysis was performed based on the model variables, the regression coefficients of the model variables were obtained and the risk factors were assigned values. A prognosis prediction model for acute myocardial infarction was constructed to screen out the risk factors that affect the occurrence of MACE events.

[0074] A nomogram model was constructed based on the acute myocardial infarction prognosis prediction model and the model was displayed.

[0075] Furthermore, the calculation formula of the acute myocardial infarction prognosis prediction model is as follows:

[0076] T=λ0+λ1×q1+λ2×q2+...+λ n ×q n

[0077] Wherein, T is the risk prediction score of MACE events after PCI for acute myocardial infarction; λ0 is the regression coefficient of the intercept of the prognostic prediction model for acute myocardial infarction; λ1, λ2 and λ n are the regression coefficients of the first, second and nth variables included in the model; q1, q2 and q n They are the 1st, 2nd and nth variables included in the model respectively.

[0078] The best, univariate regression method and multivariate regression method, due to data reasons, some important variables may not be significantly different between the two groups and cannot be included in the multivariate analysis, so usually by raising the P value to include as many variables as possible; stepwise regression includes forward regression, backward regression and bidirectional regression, among which forward regression is to include variables into the model one by one until adding variables can no longer improve the model, backward regression is to include all variables into the model and then delete variables one by one until deleting variables will reduce the quality of the model, and bidirectional regression is to first do forward regression and then backward regression; the model quality judgment criterion of regression is Akaike Information Criterion AIC, the smaller the Akaike Information Criterion AIC value, the better the model is, indicating that the model has a sufficient degree of fit with fewer parameters; optimal subset regression is a simple method to identify the optimal target model, through Malos CP value, adjusted R 2 The Bayesian Information Criterion (BIC) was used to judge the quality of the model and adjust R 2 The model constructed with the largest variable combination is optimal. The Bayesian Information Criterion BIC is similar to the Akaike Information Criterion AIC. The smaller the value, the better the model. The smaller the Malos CP value, the better the model. LASSO regression analysis selectively puts variables into the model to obtain better performance parameters, while also avoiding overfitting. Since single-factor regression method, multi-factor regression method, stepwise regression, optimal subset and LASSO regression all have the possibility of overfitting, the present invention combines the above methods to perform variable screening to minimize the occurrence of overfitting.

[0079] Specifically, for the second threshold of the P value of the judgment variable, the initial threshold is set, and variables are included. The differences of important variables in the single factor regression method and the multifactor regression method are analyzed and the threshold is adjusted to obtain the optimal second threshold.

[0080] Preferably, the performance evaluation includes model performance evaluation, model validation and model methodological quality evaluation.

[0081] Preferably, model performance evaluation includes discrimination, calibration and clinical utility.

[0082] Furthermore, the discrimination includes: evaluating using the ROC curve, if the area AUC enclosed by the ROC curve and the coordinate axis is between 0.50 and 0.70, the prediction accuracy of the acute myocardial infarction prognosis prediction model is low.

[0083] If the area under the ROC curve and the coordinate axis (AUC) is between 0.71 and 0.90, the prediction accuracy of the acute myocardial infarction prognosis prediction model is moderate.

[0084] If the area AUC enclosed by the ROC curve and the coordinate axis is higher than 0.90, the prediction accuracy of the acute myocardial infarction prognosis prediction model is high. The larger the area AUC enclosed by the ROC curve and the coordinate axis, the better the discrimination of the acute myocardial infarction prognosis prediction model. It is generally believed that the risk prediction model with an area AUC enclosed by the ROC curve and the coordinate axis greater than 0.75 has a more reliable prediction performance.

[0085] Furthermore, calibration includes Birer score, Hosmer-Lemeshow goodness-of-fit test and calibration curve.

[0086] Specifically, the closer the Bier score is to 0, the better the calibration of the acute myocardial infarction prognosis prediction model, where the value range of the Bier score is 0 to 0.25.

[0087] Specifically, the smaller the chi-square value of the Hosmer-Lemeshow goodness-of-fit test statistic, the larger the corresponding P value, and the better the calibration of the acute myocardial infarction prognosis prediction model.

[0088] Furthermore, the clinical utility includes evaluating the clinical practicality of the acute myocardial infarction prognosis prediction model using a clinical utility decision curve DCA, wherein the farther the curve in the clinical utility decision curve DCA is from the X-axis and the Y-axis, the stronger the clinical practicality of the acute myocardial infarction prognosis prediction model.

[0089] Preferably, by comprehensively evaluating the model using discrimination, calibration and clinical utility decision curve DCA, the value of the model in actual clinical application can be better judged, helping clinicians make better decisions.

[0090] Preferably, model validation includes internal validation of the acute myocardial infarction prognosis prediction model to test the repeatability of the model and prevent overfitting of the model that leads to overestimation of the model's performance.

[0091] Specifically, the internal validation includes random split validation, cross validation and Bootstrap sampling validation, wherein the present invention utilizes the internal validation methods one by one to validate the model.

[0092] Furthermore, random split validation was performed by splitting the data into a modeling group and a validation group, and performing validation in a ratio of 5:5, 6:4, 7:3, or 8:2.

[0093] Furthermore, the cross-validation was K-fold validation, which randomly divided the development cohort into 10 parts, 9 of which were used as training sets each time, and the remaining 1 was used as a validation set, and the process was repeated.

[0094] Furthermore, Bootstrap sampling verification is to randomly sample the data set and then put it back into the data set. After multiple samplings, a sampling distribution is formed to estimate the overall parameters. The number of samplings is usually 1000 times. The overall parameters include the overall mean, median, variance and quantile.

[0095] Preferably, the model methodological quality evaluation includes methodological quality evaluation using PROBAST, a tool for assessing the risk of bias and applicability of clinical prediction models, which can evaluate the risk of bias and applicability of prediction model studies. This study did not involve external validation, so only the bias risk of the model was evaluated.

[0096] Specifically, by adopting a variety of internal validation methods to evaluate the performance of the model, the evaluation accuracy of the model performance is effectively improved, and the reliability and clinical applicability of the model are enhanced.

[0097] The present embodiment also provides a system for establishing a prediction model for the prognosis of acute myocardial infarction, including: a gene screening module, which is used to collect and screen chip data sets of patients with acute myocardial infarction, mine key genes in differentially expressed genes, and obtain a first myocardial infarction key gene; a gene verification module, which is used to verify the expression level of the first myocardial infarction key gene using a myocardial infarction group and a healthy control group, and obtain a second myocardial infarction key gene; a relationship analysis module, which is used to detect patients with acute myocardial infarction, obtain the expression level of the second myocardial infarction key gene and follow up, and observe the MACE events of patients with acute myocardial infarction; a model construction module, which is used to use univariate Logistic regression analysis, stepwise regression method, optimal subset regression method, and LASSO regression method to screen variables to obtain variables included in the model, and construct an acute myocardial infarction prognosis prediction model based on multivariate Logistic regression analysis, use the Nomogram model nomogram to display the model, and evaluate the performance of the model based on model performance evaluation, model verification, and model methodology quality evaluation.

[0098] This embodiment also provides a computer device, which is suitable for a method for establishing a prediction model for the prognosis of acute myocardial infarction, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement a method for establishing a prediction model for the prognosis of acute myocardial infarction as proposed in the above embodiment.

[0099] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0100] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements a method for establishing a prediction model for the prognosis of acute myocardial infarction as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0101] In summary, the present invention, by combining immune-related LncRNA, screened out the second key gene of myocardial infarction and constructed a prognosis prediction model for acute myocardial infarction, which can effectively predict the incidence of MACE events in patients, improve the accuracy of myocardial infarction prognosis prediction, screen high-risk patients and provide more effective prevention and treatment, improve the diagnosis and treatment effect of patients with acute myocardial infarction, and reduce the waste of medical resources.

[0102] Example 2

[0103] Reference Figure 2 to Figure 17, which is the second embodiment of the present invention. This embodiment is different from the first embodiment in that, in order to verify its beneficial effects, experimental data and related instructions of the present invention in an actual environment are provided.

[0104] In this example, 210 patients with acute myocardial infarction were selected and followed up and observed. Among them, 50 patients developed MACE events, with an incidence rate of 23.8%. The specific data of the patients with acute myocardial infarction are shown in Table 1.

[0105] Table 1 Data of patients with acute myocardial infarction

[0106]

[0107] Based on the specific data of the above patients with acute myocardial infarction, univariate logistic regression analysis was used to find the factors related to MACE events. The variables with P values ​​< 0.1 included age, left atrial diameter, systolic blood pressure, Killip grade, anterior myocardial infarction, pathological Q wave, hemoglobin, total cholesterol and NEAT1. The specific data are shown in Table 2.

[0108] Table 2 Variables with P value < 0.1 in univariate logistic regression analysis

[0109] Variable Name OR (95% confidence interval) P-value age 1.05(1.02～1.08) 0.001 left atrial diameter 1.11(1.03～1.19) 0.004 Systolic blood pressure 0.2(0.05～0.75) 0.017 killip level>1 2.22(1.13～4.35) 0.021 Anterior myocardial infarction 0.5(0.26～0.96) 0.036 Hemoglobin 0.98(0.96～1) 0.05 Total cholesterol 1.25(0.97～1.6) 0.081 Pathological Q waves 0.57(0.3～1.07) 0.082 NEAT1 1.14(1.07～1.21) <0.001

[0110] After the variables were initially screened out using univariate logistic regression analysis, the model variables were determined using the backward stepwise regression method. The variables obtained included age, systolic blood pressure, anterior wall myocardial infarction, total cholesterol, left atrial diameter, and NEAT1. The backward stepwise regression started with all the predictive variables contained in the model and deleted one variable at a time until the model quality was reduced. The Akaike information criterion AIC was used to compare the goodness of fit of multiple regression models with different numbers of variables. The smaller the Akaike information criterion AIC value, the better the model performance. The variables screened by the stepwise regression are shown in Table 3, where the minimum value of the Akaike information criterion AIC is 192.015.

[0111] Table 3 Variable table of stepwise regression screening

[0112] Variable Name Estimated value Standard error Z-value P-value AIC Constant term -13.548 2.771 -4.889 0 192.015 age 0.067 0.018 3.683 0 - Systolic blood pressure -2.185 0.783 -2.79 0.005 - Anterior myocardial infarction -0.905 0.397 -2.281 0.023 - Total cholesterol 0.247 0.145 1.707 0.088 - left atrial diameter 0.095 0.046 2.071 0.038 - NEAT1 0.095 0.035 2.709 0.007 -

[0113] The optimal subset regression analysis was performed on the variables selected by univariate logistic regression analysis, including the use of adjusted R 2 , Bayesian Information Criterion BIC and Malos CP value are used to judge the quality of the model, such as Figure 2 and Figure 3 Shown is the maximum adjustment R 2 The time variables are incorporated into the process and the variables present in the model, where the maximum adjusted R 2The maximum adjusted R 2 Variables present at the time of analysis included age, left atrial diameter, systolic blood pressure, Killip grade, anterior myocardial infarction, pathological Q waves, total cholesterol, and NEAT1.

[0114] like Figure 4 and Figure 5 Shown are the variables included in the process and the variables present in the model when the minimum Bayesian information criterion BIC is obtained. The minimum Bayesian information criterion BIC is -184.6. A total of 7 variables are included. The variables present when the minimum Bayesian information criterion BIC include age, systolic blood pressure and NEAT1.

[0115] like Figure 6 and Figure 7 The figure shows the variable inclusion process and the variables present in the model when the minimum Malos CP value is 7.7. A total of 9 variables were included. The variables present when the minimum Malos CP value is 7.7 include age, left atrial diameter, systolic blood pressure, Killip grade>1, total cholesterol, and NEAT1.

[0116] Adjust R for maximum 2 The variables selected by the optimal subset regression method were unioned, including age, left atrial diameter, systolic blood pressure, Killip grade>1, anterior wall myocardial infarction, pathological Q wave, total cholesterol and NEAT1.

[0117] Table 4 Variables screened by the optimal subset regression method

[0118]

[0119] The predictive factors obtained by the above univariate logistic regression analysis were used as independent variables, and whether the outcome variable ISR occurred was used as the dependent variable for LASSO regression analysis. Figure 8 and Fig. 9 The figure shows the coefficient path diagram and cross-validation curve of the risk variables in the LASSO regression analysis. The λ value with the minimum mean square error was obtained by 10-fold cross-validation. The λ value of the simplest model was obtained within 1 variance range, that is, λ.1se=0.011 was taken as the optimal solution of the model. The model with excellent performance and the least number of independent variables was obtained. The corresponding non-zero regression coefficient variables included age, left atrial diameter, systolic blood pressure, Killip grade>1, hemoglobin, pathological Q wave, total cholesterol and NEAT1. The specific data are shown in Table 5.

[0120] Table 5 Variables screened by LASSO regression analysis

[0121] variable Lambda Types lambda value (Intercept) lambda.1se 0.059 age lambda.1se 0.059 Anterior myocardial infarction lambda.1se 0.059 Hemoglobin lambda.1se 0.059 Total cholesterol lambda.1se 0.059 left atrial diameter lambda.1se 0.059 Pathological Q waves lambda.1se 0.059 NEAT1 lambda.1se 0.059 killip level>1 lambda.1se 0.059

[0122] The variables selected by the above stepwise regression, optimal subset regression and LASSO regression analysis were intersected, and the variables finally selected for inclusion in the model included age, anterior myocardial infarction, left atrial diameter, NEAT1 and total cholesterol.

[0123] Multivariate logistic regression analysis was performed on the variables included in the above model, including age, anterior wall myocardial infarction, left atrial diameter, NEAT1 and total cholesterol. The results are shown in Table 6, in which the risk factors were assigned values ​​according to the regression coefficients in the regression model, and then the risk prediction formula for MACE events after PCI for acute myocardial infarction was obtained.

[0124] Table 6 Results of multi-factor Logistic regression analysis

[0125] Independent Variable Regression coefficient Standard error Z-value P-value (Intercept) -14.351 2.705 -5.305 0 age 0.054 0.017 3.223 0.001 Non-anterior myocardial infarction -0.97 0.388 -2.504 0.012 Total cholesterol 0.3 0.138 2.178 0.029 left atrial diameter 0.085 0.045 1.883 0.06 NEAT1 2.423 0.868 2.792 0.005

[0126] By using R language software version 3.6.3, a nomogram model nomogram for predicting MACE was drawn based on the risk ratio of risk factors. The nomogram model nomogram for the risk of MACE events in patients after PCI for acute myocardial infarction is as follows: Fig.10 As shown, the scoring lines are age 66 points, anterior wall myocardial infarction 18 points, total cholesterol 67 points, left atrial diameter 56 points and NEAT 1100 points, with a total score of 307 points.

[0127] According to the above-mentioned acute myocardial infarction prognosis prediction model, the model performance was evaluated using discrimination, such as Fig.11 Shown is the ROC curve of the acute myocardial infarction prognosis prediction model. The area under the ROC curve and the coordinate axis, AUC, is 0.794.

[0128] like Fig.12 The figure shows the calibration curve of the prediction model for the risk of ISR in patients after stent implantation. The model performance was evaluated by using the calibration degree. The Brier score of the prediction model was 0.122 points and less than 0.25, indicating that the model calibration was good and had predictive ability. The chi-square value of the Hosmer-Lemeshow goodness-of-fit test was 7.298, and the P value was 0.505 and greater than 0.05, indicating that the model fit was good. From the upper left corner of the figure, it can be seen that the model intercept is 0.00, that is, the -0.36 to 0.36 part, indicating that the difference between the observed value and the predicted value is small. The slope of the model calibration curve is 1.00, that is, the 0.64 to 1.36 part, indicating that the model calibration curve is close to the ideal curve, both indicating that the model accuracy is good.

[0129] like Fig.13 Shown is the decision curve of the prediction model for the risk of ISR in patients after stent implantation. The net benefit was calculated using the clinical decision curve to evaluate the clinical application ability of the constructed nomogram. It can be seen from the figure that the bold curve is the clinical decision curve, and the non-bold curves on both sides are the 95% confidence interval curves. The clinical decision curve analysis shows that when the threshold probability of predicting ISR is in the range of 1% to 81%, the practicality of using the nomogram to predict ISR is best.

[0130] To verify the internal validity of the model, multiple validations were performed using random split validation, cross validation, and bootstrap sampling validation. Fig.14 and Fig.15 The ROC curve and calibration curve of random split validation are shown. The model is validated by dividing the samples in a ratio of 8:2 between training set and validation set, that is, 80% of the data is used to build the model and 20% of the data is used to validate the model. From the ROC curve of the random split validation set, it can be seen that the area under the ROC curve and the coordinate axis, AUC, is 0.797, the 95% confidence interval is 0.720-0.874, the specificity is 73.4%, and the sensitivity is 67.0%, indicating that the model has good discrimination. From the calibration curve of the random split validation set, it can be seen that the Brier score is 13.9, the 95% confidence interval is 10.7-17.1, the Hosmer-Lemeshow goodness of fit test chi-square value is 5.171, and the P value is 0.739 and greater than 0.05, indicating that the calibration is good.

[0131] Table 7 shows a comparison of the key indicators of the original model and the cross-validation model. By using the cross-validation method, i.e., K-fold validation, the development cohort is randomly divided into 10 parts, 9 of which are used as training sets each time, and the remaining 1 is used as a validation set, and this process is repeated. The corresponding cross-validation model calibration curve is shown in Fig.16 As shown, it can be seen that there is no statistically significant difference in the main evaluation indicators of the model, C index, Brier score, and between the cross-validation mean group and the original value group, with P value > 0.05.

[0132] Table 7 Comparison of key indicators between the original model and the cross-validation model

[0133] project Original value Cross-validation mean P C(ROC) 0.794 0.769 ＞0.05 Brier 0.143 0.154 ＞0.05

[0134] According to the TRIPOD guidelines, the Bootstrap method is used to evaluate the performance of the model, which is considered to be the most effective model validation method. To prevent overfitting, Bootstrap self-sampling validation is used, such as Fig.17The figure shows the Bootstrap sampling validation calibration curve. It can be seen from the figure that the adjusted AUC obtained after 1000 bootstrap samplings is 0.733, which is not much different from the original model AUC and is basically consistent with the actual curve.

[0135] Regarding the evaluation of the methodological quality of the model, since this study did not involve external validation, only the risk of bias of the model was evaluated. After rigorous review using the PROBAST tool, the final overall evaluation result of the model was a low risk of bias.

[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for establishing a prediction model for the prognosis of acute myocardial infarction, characterized in that: include: Collect and screen the patient chip data to obtain the first key gene for myocardial infarction; Verifying the first myocardial infarction key gene to obtain a second myocardial infarction key gene; Based on the second myocardial infarction key gene, the relationship between the key gene expression level and the patient's prognosis was analyzed; The relationship between the expression level of the key genes and the patient's prognosis was used to construct an acute myocardial infarction prognosis prediction model and conduct performance evaluation.

2. The method for establishing a prediction model for the prognosis of acute myocardial infarction according to claim 1, characterized in that: The first myocardial infarction key gene comprises the following steps: Download the chip dataset of acute myocardial infarction patients based on the same platform from the database; Performing background correction and quantile normalization on the chip data set of patients with acute myocardial infarction, and plotting the corrected data; The corrected data were subjected to differential expression analysis to obtain differentially expressed genes; Perform GO analysis and KEGG pathway enrichment analysis on differentially expressed genes, predict protein interaction PPI network, and mine key genes among differentially expressed genes; The M key genes before myocardial infarction were screened out to obtain the first key gene for myocardial infarction.

3. The method for establishing a prediction model for the prognosis of acute myocardial infarction as claimed in claim 2, characterized in that: The second myocardial infarction key gene comprises the following steps: A patient with acute myocardial infarction was selected and divided into the myocardial infarction group; A patient with normal coronary angiography was selected and divided into the healthy control group; The RT-qPCR technology was used to detect the expression level of the first myocardial infarction key gene in the myocardial infarction group and the healthy control group, that is, the level of plasma LncRNA was detected, and the top N key genes consistent with the actual clinical situation were screened out to obtain the second myocardial infarction key gene.

4. The method for establishing a prediction model for the prognosis of acute myocardial infarction as claimed in claim 3, characterized in that: The analysis of the relationship between the key gene expression level and the patient's prognosis comprises the following steps: The RT-qPCR technique was used to detect the acute myocardial infarction patient B, and the expression level of the second myocardial infarction key gene was obtained; The patient B with acute myocardial infarction was followed up for C years to observe the patient's adverse cardiovascular events, namely MACE events; The expression levels of the key genes for the second myocardial infarction were combined with the patient's adverse cardiovascular events to obtain the relationship between the expression levels of the key genes and the patient's prognosis.

5. The method for establishing a prediction model for the prognosis of acute myocardial infarction as claimed in claim 4, characterized in that: The MACE events included angina attack, recurrent acute myocardial infarction, target vessel revascularization, worsening cardiac function and all-cause mortality.

6. The method for establishing a prediction model for the prognosis of acute myocardial infarction according to claim 5, characterized in that: The construction of the acute myocardial infarction prognosis prediction model comprises the following steps: Based on the Harrell guidelines for sample size in multivariate analysis, the sample size was collected according to the rule that the number of event occurrences EPV> the first threshold; Based on the sample size, a univariate logistic regression analysis is performed to obtain the P value of the variable and make a judgment. If the P value of the variable is less than the second threshold, the corresponding variable is determined to be associated with the MACE event and recorded to obtain the first predictor variable. If the P value of the variable is ≥ the second threshold, the corresponding variable is judged to be unrelated to the MACE event; The first predictor variable is analyzed by using a stepwise regression method to obtain the Akaike information criterion AIC of the first predictor variable and make a judgment. If the Akaike information criterion AIC of the first predictor variable reaches the minimum, the corresponding variable is classified as the second predictor variable; The first predictor variable is analyzed using the best subset regression method to obtain the adjusted R of the first predictor variable. 2 , Bayesian Information Criterion BIC and Malos CP value and make judgments. If the adjusted R 2 reaches the maximum, the corresponding variable is divided into the maximum adjustment R 2 predictor variables; If the Bayesian Information Criterion (BIC) of the first predictor variable reaches the minimum, the corresponding variable that exists is classified as the predictor variable with the minimum Bayesian Information Criterion (BIC); If the Malos CP value of the first predictor variable reaches the minimum, the corresponding variable that exists is classified as the predictor variable with the minimum Malos CP value; Based on the maximum adjustment R 2 The predicted variable of the minimum Bayesian information criterion BIC and the predicted variable of the minimum Malos CP value are taken as a union to obtain a third predicted variable; The first prediction variable is analyzed by using the LASSO regression method to obtain the lambda value of the mean square error and make a judgment. If the lambda value of the mean square error reaches the minimum, it is determined that the number of independent variables is the minimum, and the corresponding non-zero regression coefficient variable is divided into the fourth prediction variable; Taking the intersection of the second predictor variable, the third predictor variable and the fourth predictor variable to obtain a variable to be incorporated into the model; Based on the model variables, multi-factor Logistic regression analysis is performed to obtain regression coefficients of the model variables and assign values ​​to risk factors to construct an acute myocardial infarction prognosis prediction model; A nomogram model was constructed based on the acute myocardial infarction prognosis prediction model and the model was displayed.

7. The method for establishing a prediction model for the prognosis of acute myocardial infarction according to claim 6, characterized in that: The calculation formula of the acute myocardial infarction prognosis prediction model is as follows: T=λ0+λ1×q1+λ2×q2+...+λ n ×q n Wherein, T is the risk prediction score of MACE events after PCI for acute myocardial infarction; λ0 is the regression coefficient of the intercept of the prognostic prediction model for acute myocardial infarction; λ1, λ2 and λ n are the regression coefficients of the first, second and nth variables included in the model; q1, q2 and q n They are the 1st, 2nd and nth variables included in the model respectively.

8. A system for establishing a prediction model for the prognosis of acute myocardial infarction, based on the method for establishing a prediction model for the prognosis of acute myocardial infarction according to any one of claims 1 to 7, characterized in that: include, Gene screening module, used to collect and screen the chip data set of patients with acute myocardial infarction, mine the key genes among the differentially expressed genes, and obtain the first key gene of myocardial infarction; A gene verification module is used to verify the expression level of the first myocardial infarction key gene using the myocardial infarction group and the healthy control group to obtain the second myocardial infarction key gene; The relationship analysis module is used to detect patients with acute myocardial infarction, obtain the expression level of the second myocardial infarction key gene and conduct follow-up to observe the MACE events of patients with acute myocardial infarction; The model building module is used to screen variables using univariate logistic regression analysis, stepwise regression method, optimal subset regression method, and LASSO regression method to obtain model variables, and to build an acute myocardial infarction prognosis prediction model based on multivariate logistic regression analysis. The Nomogram model is used to display the model, and the performance of the model is evaluated based on model performance evaluation, model verification, and model methodological quality evaluation.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for establishing a prediction model for the prognosis of acute myocardial infarction as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for establishing a prediction model for the prognosis of acute myocardial infarction as described in any one of claims 1 to 7 are implemented.

Images