【摘要】 我国心血管病负担日渐加重,其中急性心肌梗塞(AMI)患病率近年来有上升趋势,防治心血管病刻不容缓。对AMI的患病机理进行了研究,并对AMI患病可能性进行预测。基于涵盖行为数据、人口学指标等的10万名患者2016年全年数据,利用机器学习算法,构建了对2017年第1季度患者是否会患AMI的预测模型。并对比了利用不同算法,包括决策树、逻辑回归、随机森林以及GBDT算法,进行预测的准确性及泛化能力。研究结果表明,结合患者行为数据进行AMI患病风险预测是可行的,且决策树及随机森林算法可达到较高的预测准确性。研究成果对于提高对AMI患病风险的预测能力,更好的服务于AMI的防病管理有一定的借鉴意义,并可应用于保险行业。更多还原

【Abstract】 With the increasing burden of cardiovascular disease and the prevalence of Acute Myocardial Infarction(AMI),it is urgent to prevent cardiovascular disease. This article studies the pathogenesis of AMI and aims to predict the risk of AMI. Based on the 2016 annual data of 100,000 patients covering behavioral data and demographic indicators, machine learning algorithms were used to construct a predictive model for patients with AMI in the first quarter of 2017. The accuracy and generalization ability of prediction were compared by using different algorithms, including Decision Tree, Logistic Regression, Random Forest and GBDT algorithm. The results show that it is feasible to predict the risk of AMI disease by combining patient behavior data, and the Decision Tree and Random Forest algorithm can achieve high prediction accuracy. The research results have certain reference significance for improving the predictive ability of AMI risk and better serving the disease prevention management of AMI,and can be applied to the insurance industry.更多还原