大数据背景下医院门诊挂号预约爽约行为预测研究  

An Investigation of Predicting Patient Missing Appointment Behavior Under the Big Data Background

作　　者：朱光 邓弘林 ZHU Guang;DENG Hong-lin(Department of Publicity and Planning,Cangzhou Central Hospital,Cangzhou 061000,Hebei,China;School of Business,Sun Yat-sen University,Guangzhou 510000,Guangdong,China)

机构地区：[1]沧州市中心医院宣传策划部,河北沧州061000 [2]中山大学管理学院,广东广州510000

出　　处：《医学信息》2020年第22期13-15,21,共4页Journal of Medical Information

摘　　要：目的了解目前医院预约诊疗服务中患者爽约行为的现状,探讨和鉴别患者爽约的关键特征,运用这些特征建立机器学习算法模型预测未来患者爽约行为。方法挖掘2018年河北省某大型三甲医院预约大数据,首先用Stata采取传统Logistic回归找出患者爽约的显著因子,再将数据划分为训练集和预测集,采用SVM、决策树、随机森林和BP神经网络等不同模型学习训练患者爽约行为和特征,检验每种算法对患者爽约预测的准确率。结果目前医院患者预约爽约率为16.16%,Logistic回归分析显示年龄、性别、预约时间和预约科室是爽约行为的关键性特征;使用这些特征进行机器学习和预测能取得较好效果,SVM、决策树、随机森林和BP神经网络各个算法准确率均超过75%,其中SVM和BP神经网络准确率最高,是该特定情境下的最优算法。结论我国大型三甲医院预约诊疗服务有待进一步加强,在大数据时代的背景下,机器学习方法可为医院预测并降低爽约率提供强有力支持。Objectives To understand the current situation of patients'missing-appointment behavior in the appointment service of hospitals;to explore and identify the key features of patients'missing appointment.Use these features to build a machine learning algorithm model to predict future patient missing-appointment behavior.Methods Mining the big data of appointments in a large tertiary hospital in Hebei Province in 2018.First,Stata adopts traditional Logistic regression to find the significant factors of patients'appointments,and then divides the data into training sets and prediction sets,using SVM,decision tree,random forest and BP Different models,such as neural networks,learn and train patients'absentee behavior and characteristics,and test the accuracy of each algorithm in predicting patient absenteeism.Results The current appointment rate of hospital patients is 16.16%.Logistic regression analysis shows that age,gender,appointment time and appointment department are the key features of appointment cancellation behavior;using these features for machine learning and prediction can achieve better results,SVM,decision tree accuracy of each algorithm of random forest and BP neural network exceeds 75%.Among them,SVM and BP neural network have the highest accuracy,which is the best algorithm in this specific situation.Conclusion The appointment diagnosis and treatment services of my country's large tertiary hospitals need to be further strengthened.In the context of the era of big data,machine learning methods can provide strong support for hospitals to predict and reduce the rate of missing-appointment.

关 键 词：预约诊疗 爽约率 机器学习 大数据 

分 类 号：R197.3[医药卫生—卫生事业管理]