先天性上睑下垂住院患者特征分析及人数预测  

Characteristics Analysis and Number Prediction of Hospitalized Patients with Congenital Ptosis

作　　者：王慧[1] 宋雪霏 杨辰玲 王一 李凌子 周慧芳[1] 孙静[1] WANG Hui;SONG Xue-fei;YANG Chen-ling;WANG Yi;LI Ling-zi;ZHOU Hui-fang;SUN Jing(Department of Ophthalmology,Shanghai Ninth People’s Hospital,Shanghai Jiao Tong University School of Medicine/Shanghai Key Laboratory of Orbital Diseases and Ocular Oncology,Shanghai 200011,China)

机构地区：[1]上海交通大学医学院附属第九人民医院眼科/上海市眼眶病眼肿瘤重点实验室,上海200011

出　　处：《医学信息》2023年第13期14-18,23,共6页Journal of Medical Information

基　　金：上海交通大学医学院附属第九人民医院样本库基金学科特色项目(编号:YBKB201901)。

摘　　要：目的通过对住院病案首页数据进行分析,了解先天性上睑下垂住院患者基本情况,探索自回归综合移动平均(ARIMA)乘积季节模型预测住院人数的适用性,为科室合理安排医务人员和医疗资源提供依据。方法以2013年1月-2019年12月我院眼科先天性上睑下垂患者病案首页数据为基础,描述性分析患者基本情况,绘制患者地区分布条形图,建立住院人数的ARIMA乘积季节模型,计算平均绝对百分误差(MAPE)和平均相对误差(MRE),检验模型的预测能力。结果2013-2019年共收治先天性上睑下垂患者3646例,患者平均年龄(6.68±5.28)岁,源自国内外30个地区、15个民族,平均住院天数(3.91±1.63)d,平均住院总费用(1.06±0.39)万元。通过R语言auto.arima语句,自动确定最优模型为ARIMA(1,0,1)×(0,1,1)_(12),AIC=478.94,Ljung-Box Q=4.738,P=0.578,并且该模型的所有参数均通过统计学检验(P<0.001);2013年1月-2019年6月拟合值与实际值之间的MAPE为11.92,2019年7月-12月预测值与实际值之间的MRE为19.95%。结论ARIMA乘积季节模型可基本拟合先天性上睑下垂住院人数的演变趋势,今后可应用于科室住院人数、门诊人数、医疗器械使用量等的预测。Objective To understand the basic situation of hospitalized patients with congenital ptosis by analyzing the data on the medical record home page,and to explore the applicability of the product seasonal model of auto-regressive integrated moving averages(ARIMA)to predict the number of hospitalized patients,so as to provide a basis for the reasonable arrangement of medical staff and medical resources.Methods Based on the data of the medical record home page of patients with congenital ptosis in ophthalmology department of our hospital from January 2013 to December 2019,the basic situation of patients was descriptively analyzed,the bar chart of regional distribution of patients was drawn,and the ARIMA product seasonal model of the number of inpatients was established.The mean absolute percentage error(MAPE)and mean relative error(MRE)were calculated to test the predictive ability of the model.Results From 2013 to 2019,3646 patients with congenital ptosis were treated,with an average age of(6.68±5.28)years,from 30 regions and 15 nationalities at home and abroad.The average length of hospitalization was(3.91±1.63)days,and the average total hospitalization cost was(1.06±0.39)ten thousand yuan.The optimal model was automatically determined as ARIMA(1,0,1)×(0,1,1)_(12) through R language auto.arima statement,AIC=478.94,Ljung-Box Q=4.738,P=0.578.And all the parameters of the model passed the statistical test(P<0.001).The MAPE between the fitted value and the actual value from January 2013 to June 2019 was 11.92,and the MRE between the predicted value and the actual value from July 2019 to December 2019 was 19.95%.Conclusion The ARIMA multiplicative seasonal model can basically fit the evolution trend of congenital ptosis hospitalization number,and can be applied to the prediction of department hospitalization number,outpatient number,medical device use,etc.

关 键 词：ARIMA乘积季节模型 住院量 病案统计 先天性上睑下垂 

分 类 号：R777.1[医药卫生—眼科]