Construction of a predictive model for acute myocardial infarction based on LASSO regression in elderly patients with acute coronary syndrome after PCI

Song Bochao; Zhu Lei; Xiong Min

doi:10.3969/j.issn.1674-4055.2025.06.12

Abstract：Objective To screen the factors for complicated acute myocardial infarction(AMI)based on LASSO model in elderly patients with acute coronary syndrome(ACS)after percutaneous coronary intervention(PCI),and construct a monogram model based on these influence factors.Methods ACS patients(n=166)were selected from Department of Cardiovascular Medicine in People's Hospital of Shijiazhuang City from Aug.2022 to Mar.2024 through a prospective study.According to AMI occurrence,the patients were divided into AMI group(n=56)and non-AMI group(n=110).The baseline data and laboratory indicators were compared between 2 groups.The factors for AMI occurrence were screened by using LASSO regression in elderly ACS patients after PCI,and these factors were analyzed by using binary Logistic regression.a nomogram model was constructed based on the analysis results and internal validation was conducted.The predictive value of the model for AMI occurrence was analyzed by using decision curve analysis in elderly ACS patients after PCI.Results Among 166 ACS patients,56 complicated by AMI and incidence rate was 33.73%.The number of stents,diastolic blood pressure(DBP),systolic blood pressure(SBP),left ventricular end-diastolic diameter(LVEDD),platelet to lymphocyte ratio(PLR)and fibrinogen(FIB)were higher,and left ventricular ejection fraction(LVEF),albumin(ALB),multi-vessel disease and percentage of postoperative intensive statin treatment were lower in AMI group than those in non-AMI group(P＜0.05).The optimal penalty coefficient λ was determined through three-fold cross-validation of LASSO regression model.There were 10 potential relevant factors screened at this λ value ultimately,including postoperative intensive statin therapy,number of stents,multi-vessel disease,SBP,DBP,LVEDD,PLR,FIB,LVEF and ALB.The results of binary Logistic regression analysis showed that number of stents,multi-vessel disease,SBP,DBP,PLR and FIB were independent risk factors(OR＞1,P＜0.05),and ALB was a protective factor(OR＜1,P＜0.05)for AMI occurrence in elderly ACS patients after PCI.The internal validation of the nomogram model using the Bootstrap method showed a C-index value of 0.992,indicating that the model has excellent discriminative ability.The results of ROC curve analysis showed that,in predicting AMI occurrence after PCI,AUC of the nomogram model was 0.992(95%CI:0.983～1.000,P＜0.001),specificity was 0.973,sensitivity was 0.946,and Youden index was 0.919.The results of decision curve analysis showed that,within the threshold range of 0.000 to 1.000,the net benefit of combination of was superior to that of each indicator alone in predicting AMI occurrence in elderly ACS patients after PCI.Conclusion The influence factors for AMI occurrence in elderly ACS patients after PCI including number of stents,number of diseased vessels,DBP,SBP,PLR,FIB and ALB.The monogram model constructed based on above factors can effectively predict AMI risk in elderly ACS patients after PCI.