AMI被视为威胁人类生命的第一杀手，而PCI作为治疗AMI的有效手段，在显著降低其病死率的同时，仍无法避免术后MACE的发生；如何有效预测和防止MACE是改善AMI患者预后的关键。鉴于此，本研究将机器学习模型与临床双向队列研究相结合，利用数据挖掘技术和WRF模型的自学习能力，深度挖掘生物遗传、行为方式、临床治疗等多元数据与疾病进展的关系，构建的WRF模型可最大限度的从数据中获取预后影响因素并整合，用于AMI患者PCI术后发生MACE的风险预测。通过计算和分析精确率、ROC-AUC、ACC等模型评估指标，并与传统的ANN、Cox回归模型预测效果进行对比，同时，结合决策分析技术优选预测模型。该预测模型一旦建立，可评价、预测和提前对PCI手术患者进行干预，降低MACE的发生，提高患者的生存质量和生存率，丰富AMI防治理论与方法。

AMI is regarded as the first killer to threaten human life, and PCI significantly prevent postoperative MACE whereby reducing mortality. It is the key of improving the prognosis of AMI, which can effectively predict and prevent MACE. In view of this, this study combines the machine learning model with the clinical two-way cohort study, and uses the self-learning ability of data mining technology and WRF model to deeply explore the relationship between family history, lifestyle, surgery, drug therapy and other multivariate data and disease progression.The WRF model can maximize the prognostic factors from the data and integrate them for prognosis prediction and risk assessment after PCI in patients with AMI. At the same time,through the accuracy evaluation, ROC-AUC, ACC and other model evaluation indicators, compared with the traditional ANN, Cox regression model prediction results, combined with decision analysis techniques to optimize the best model. The predictive model can evaluate the intervention of patients after PCI, reduce the occurrence of MACE, improve the quality of life and survival rate of patients, and enrich the theory and method of AMI prevention.