基于数据挖掘技术的冠心病诊断预测模型  

Diagnosis and Prediction Model of Coronary Heart Disease Based on Data Mining Technology

作　　者：李雨洁 郑锐龙 杨旭明[1] LI Yu-jie;ZHENG Rui-long;YANG Xu-ming(School of Acupuncture and Tuina,Shanghai University of Traditional Chinese Medicine,Shanghai 201203,China)

机构地区：[1]上海中医药大学针灸推拿学院,上海201203

出　　处：《医学信息》2020年第24期14-17,共4页Journal of Medical Information

基　　金：上海中医药大学第十三批大学生创新创业活动计划资助项目(编号:2020SHUTCM067)。

摘　　要：目的通过WEKA数据挖掘平台将数据挖掘技术运用在冠心病的诊断中,寻找最佳的冠心病诊断预测模型。方法从UCI数据库中的272个冠心病实例中筛选出259例患者的数据进行分析,采用决策树算法、MultilayerPerceptron算法、关联规则分析分别获取影响冠心病诊断结果的规则。结果通过J48决策树获得了具有一定指导意义决策模型,能为冠心病的诊断提供辅助决策支持;通过Apriori算法,借助信息增益的属性约简,获取到10条正确率相对较高的规则;根据MultilayerPerceptron算法,得出关于12种输入属性对于冠心病诊断量化的影响力。结论使用WEKA数据挖掘平台建立的冠心病早期预测模型对于帮助医护人员判断病情具有一定的指导意义。Objective To apply data mining technology to the diagnosis of coronary heart disease through the WEKA data mining platform,and to find the best prediction model for coronary heart disease diagnosis.Methods Data of 259 patients were selected from 272 cases of coronary heart disease in the UCI database for analysis,and the decision tree algorithm,MultilayerPerceptron algorithm,and association rule analysis were used to obtain the rules affecting the diagnosis of coronary heart disease.Results A decision model with certain guiding significance was obtained through the J48 decision tree,which could provide auxiliary decision support for the diagnosis of coronary heart disease;through the Apriori algorithm,with the help of attribute reduction of information gain,10 rules with relatively high accuracy rates were obtained;MultilayerPerceptron algorithm,obtained about the influence of 12 input attributes on the quantification of coronary heart disease diagnosis.Conclusion The early prediction model of coronary heart disease established by WEKA data mining platform has certain guiding significance for helping medical staff to judge the condition.

关 键 词：J48决策树 关联规则 数据挖掘 WEKA 人工神经网络 

分 类 号：R541.4[医药卫生—心血管疾病] TP311.1[医药卫生—内科学]