摘要
采用随机森林算法建立糖尿病预测模型,介绍数据建模与评估步骤和方法、选择标准。以受试者工作特征曲线的曲线下面积、灵敏度、特异性、正确率等指标评价模型的预测效果,结果表明随机森林预测模型对糖尿病具有较强的预测能力。
The prediction models of diabetes mellitus has been built by taking random forest algorithm.The paper introduces data modeling,steps and methods of evaluation,selection criteria,and evaluates the predictive effect of the models using indicators of the area under curve of receiver operating characteristic curve,sensitivity,specificity,accuracy,etc.The result indicates that the random forest prediction model possesses a relative strong prediction ability as for diabetes mellitus.
作者
杨美洁
唐建军
YANG Meijie;TANG Jianjun(Medical Informatics College,Chongqing Medical University,Chongqing 400016,China)
出处
《医学信息学杂志》
CAS
2019年第9期47-49,共3页
Journal of Medical Informatics
基金
重庆市社会事业与民生保障科技创新专项(项目编号:cstc2015shms-ztzx10003)
关键词
随机森林
糖尿病
灵敏度
特异性
AUC
random forest
diabetes mellitus
sensitivity
specificity
AUC