Abstract: Background　Acute myocardial infarction（AMI） is one of the leading causes of death worldwide，and percutaneous coronary intervention（PCI） is the preferred clinical treatment option for most AMI patients. However，patients still face a high risk of heart failure（HF） post-PCI，with poor prognosis，severely impacting their quality of life and long-term survival rates. Objective　The risk prediction model of HF after PCI in patients with AMI was evaluated with the aim of providing reference for the development，optimisation and application of the model. Methods　A computerized systematic search was conducted across PubMed，Web of Science，Embase，CNKI，VIP and Wanfang Data for relevant literature on HF risk prediction models in AMI patients post-PCI，with the search period spanning from database inception to November 12，2024. Two researchers independently screened the literature，extracted data，and assessed the quality of included studies using the PROBAST tool. Meta-analysis of HF risk predictors in AMI patients post-PCI was performed using RevMan 5.4 software. Results　A total of 18 papers were included，with a sample number of 6 375 cases，and the incidence rate of HF was 13%. The incidence of HF ranged from 13.7% to 34.8%，with a total of 21 HF risk prediction models for post-PCI in patients with AMI，and the area under the curve of the working characteristic（AUC） of the subjects ranged from 0.657% to 0.966%，with 17 prediction models having good predictive performance（AUC>0.7），and the overall bias of the literature was high risk of bias and good applicability. Statistical analysis of model predictive performance AUC was performed using MedCalc software，and the combined AUC was 0.852（95%CI=0.815-0.890）. Meta-analysis results indicated that age，Gensini score，arrhythmia，serum creatinine（Scr），ventricular wall motion amplitude，hypertension，diabetes mellitus，left ventricular ejection fraction （LVEF），high-sensitivity C-reactive protein（hsCRP），N-terminal pro-brain natriuretic peptide（NT-proBNP），brain natriuretic peptide（BNP），number of lesion vessels，cardiac structural changes，and anterior wall myocardial infarction were predictors of HF in AMI patients post-PCI（P<0.05）. Conclusion　Currently，the risk prediction risk prediction model for post-PCI HF risk prediction in AMI patients is still in the exploratory stage，with good model differentiation but high overall risk of bias，and the study design should be optimised and the reporting process should be improved in the future in order to ensure that a predictive model with strong clinical utility is developed.

Key words: Acute myocardial infarction, Heart failure, Prediction model, Systematic evaluation, Meta-analysis

摘要： 背景　急性心肌梗死（AMI）是全球主要的死亡原因之一，经皮冠状动脉介入术（PCI）是多数AMI患者临床治疗的首选方式，但是术后仍面临较高的心力衰竭（HF）的风险，且预后较差，严重影响患者的生存质量和长期生存率。目的　对AMI患者PCI术后HF的风险预测模型进行评估，为模型的开发、优化和应用提供参考。方法　计算机系统检索Pubmed、Web of Science、Embase、中国知网（CNKI）、维普网（VIP）和万方数据知识服务平台数据库中关于AMI患者PCI术后HF风险预测模型的相关文献，检索时限为数据库创建至2024-11-12，由2位研究员独立筛选文献、提取数据并采用PROBAST评价纳入文献的质量，应用RevMan 5.4软件对AMI患者PCI术后HF风险预测因素进行Meta分析。结果　共纳入18篇文献，样本量6 375例，HF发生率为13.7%~34.8%，共21个AMI患者PCI术后HF风险预测模型，受试者工作特征（ROC）曲线下面积（AUC）为0.657%~0.966%，有17个预测模型的预测性能较好（AUC>0.7），文献总体偏倚均为高偏倚风险，适用性好。采用MedCalc软件对模型预测性能AUC进行统计学分析，合并AUC为0.852（95%CI=0.815~0.890）。Meta分析结果显示：年龄、Gensini评分、心律失常、血清肌酐（Scr）、室壁运动幅度、高血压、糖尿病、左室射血分数（LVEF）、高敏C反应蛋白（hsCRP）、氨基末端脑钠肽前体（NT-proBNP）、脑钠肽（BNP）、多病变支数、心脏结构改变、前壁心肌梗死是AMI患者PCI术后HF的预测因子（P<0.05）。结论　目前AMI患者PCI术后HF预测模型仍处于探索阶段，模型区分度较好，但总体偏倚风险较高，未来应优化研究设计和完善报告流程，以确保开发出临床实用性强的预测模型。

关键词: 急性心肌梗死, 心衰预测, 预测模型, 系统评价, Meta 分析