差分整合移动平均自回归模型乘积季节模型在病毒性肝炎门诊量预测中的应用

Application of ARIMA product seasonal model in the prediction of number of viral hepatitis outpatients

目的分析某三级综合医院病毒性肝炎门诊量的变化并建立合适的模型,预测其就诊量变化趋势,为医院决策提供依据。方法运用Excel 2019软件建立数据库,SPSS 22.0软件对2005—2018年病毒性肝炎门诊量数据进行建模,2019年数据进行模型验证。结果病毒性肝炎门诊量整体呈下降趋势,但2017年后有回升趋势。采用传统建模方法和专家建模器拟合最优模型均为差分整合移动平均自回归模型(ARIMA)(0,1,1)(1,0,1)12。模型残差检验显示,残差均为白噪声序列,经典建模和专家建模器所建模型各项指标相似,平稳的R~2均为0.336,标准化的BIC值分别为6.126、6.089。2019年预测数据显示短期预测效果较好,而长期预测效果不理想。结论采用专家建模器构建的乘积季节模型在病毒性肝炎门诊量短期预测中预测效果较好,该方法客观、高效、简单,可用于门诊量短期预测。

Objective To construct the optimal model and to predict the trend of viral hepatitis outpatients by analyzing the number of previous outpatients in a gradeⅢclass A hospital,which would provide a basis for the decision-making.Methods The number of viral hepatitis outpatients in a gradeⅢclass A hospital from 2005 to 2018 were summarized by Excel 2019 software,and the model used to predict number of viral hepatitis outpatients was constructed by using SPSS 22.0 software,by which the prediction effect was then validated by using 2019 data.Results From 2005 to 2018,the viral hepatitis outpatients decreased yearly,but showed a rising trend after 2017.ARIMA(0,1,1)(1,0,1)12 was determined as the optimal model fitted by both the traditional modeling method and the expert modeler.The residual test of the model showed that the residual were white noise sequences,and the indexes of the model built by the classical model and the expert modeler were similar.The stable R2 was 0.336,and the standardized BIC value was 6.126 and 6.089,respectively.After validation by using the 2019 data,it was shown that the effect of short-term prediction was better than that of the long-term prediction.Conclusion The product seasonal model constructed by the expert modeler is objective,efficient and simple,which could be used in short-term prediction of number of viral hepatitis outpatients.