自动回归移动平均混合模型对重症肺炎患者日均住院费用月份变化趋势分析及预测价值  

Regression analysis and prediction value of daily hospitalization expenses monthly variation of severe pneumonia patients by auto regressive moving average mixture model

作　　者：官和立[1] 颜晓玉[1] 昝海峰[1] 熊波[1] 周晖[1] 

机构地区：[1]成都大学附属医院呼吸内科和肺血管病诊治中心,四川省成都市610081

出　　处：《临床合理用药杂志》2017年第18期101-103,共3页Chinese Journal of Clinical Rational Drug Use

基　　金：四川省卫生与计划生育委员会研究项目(NO.140001);成都大学校青年基金(自然科学类)重点项目(NO.2015XJZ12);四川基层卫生事业发展中心研究项目(NO.SJWF1423)

摘　　要：目的探讨自动回归移动平均混合模型(ARIMA)对重症肺炎患者日均住院费用月份变化趋势拟合及预测价值。方法回顾性分析成都大学附属医院2013年7月—2015年6月重症肺炎患者80例,以月份为单位,采用ARIMA模型对患者日住院费用数据进行分析和预测,用Box-Ljung法分析平均预测相对误差作为预测效果的评价指标。结果重症肺炎患者日均住院费用数据进行平稳化检验、差分、模型识别、获得ARIMA(0,1,0)为最适模型,拟合效果检验显示,预测值与实际值吻合程度较高,模型拟合效果良好。结论 ARIMA模型可较好地分析和预测重症肺炎患者日均住院费变化趋势。Objective To explore the prediction value of daily hospitalization expenses monthly variation of severe pneumonia patients by auto regressive moving average mixture model（ARIMA）.Methods 80 cases of patients with severe pneumonia were retrospective analysized from July 2013 to June 2015 in Affiliated Hospital of Chengdu University,took month as a unit,the daily hospitalization expenses was analysized and predicted by ARIMA model,and the average relative prediction error was as the evaluation index to predict the effect by using Box-Ljung method.Results The hospitalization costs data of severe pneumonia patients was given stationary test,differential,model identification,ARIMA（0,1,0） was the most suitable model,and the fitting effect tests showed that the predicted values were in accordance with the actual value of higher degree,the model fitting effect was good.Conclusion ARIMA model can preferably analyze and predict the trend of daily hospitalization expenses of severe pneumonia patients.

关 键 词：肺炎 ARIMA模型 日均住院费用 时间序列 

分 类 号：R563.1[医药卫生—呼吸系统]