首都医科大学学报 ›› 2019, Vol. 40 ›› Issue (2): 286-291.doi: 10.3969/j.issn.1006-7795.2019.02.024

• 基础研究 • 上一篇    下一篇

差分整合移动平均自回归模型在医院流感样病例监测中的应用

李桂芹1, 黄立勇2, 覃凤芝3   

  1. 1. 首都医科大学附属北京朝阳医院疾病预防控制处, 北京 100020;
    2. 北京市朝阳区疾病预防控制中心流行病与地方病控制科, 北京 100021;
    3. 厦门大学公共卫生学院, 厦门 361102
  • 收稿日期:2018-11-20 出版日期:2019-03-21 发布日期:2019-04-15
  • 通讯作者: 李桂芹 E-mail:liguiqinlgq@163.com

Application of autoregressive integrated moving average model on monitoring influenza-like illness(ILI) case in hospital

Li Guiqin1, Huang Liyong2, Qin Fengzhi3   

  1. 1. Office for Disease Prevention and Control, Chaoyang Hospital, Capital Medical University, Beijing 100020, China;
    2. Department of Infectious Disease and Endemic Disease Prevention, Chaoyang Centers For Disease Control and Prevention, Beijing 100021, China;
    3. College of Public Health, Xiamen University, Xiamen 361102, Fujian Province, China
  • Received:2018-11-20 Online:2019-03-21 Published:2019-04-15

摘要: 目的 应用差分整合移动平均自回归模型(autoregressive integrated moving average model,ARIMA)分析医院流感样病例报告数据,初步探索ARIMA模型在流感样病例监测和预警上的效果,以期更好地指导医院相关医务人员应对秋冬季流感就诊高峰,全面开展流感防治工作,及时有效地应对疫情。方法 利用2014年1月12日至2017年10月14日间首都医科大学附属北京朝阳医院每日报告的流感样病例数据建立ARIMA模型,选取2017年10月15日至12月24日的流感样病例数据作为检验集来评价模型。结果 ARIMA(2,0,0)模型应用于首都医科大学附属北京朝阳医院流感样病例时,决定系数(R2)为0.87。用该模型进行回代预测,预测值与实际值吻合程度较高。结论 ARIMA(2,0,0)模型分析结果显示该模型在首都医科大学附属北京朝阳医院流感预测中具有较好的效果。可为其他医疗机构在流感样病例监测工作中提供借鉴依据。

关键词: 流感样病例, 监测, 时间序列模型

Abstract: Objective To analyze the influenza-like illness (ILI) cases data reported via autoregressive integrated moving average model (ARIMA) model, and to explore the effect of the model on ILI case warning, to carry out the influenza prevention work comprehensively, and to deal with the epidemic situation efficiently.Methods The ARIMA model was established via the ILI cases data reported by Chaoyang Hospital, Capital Medical University from January 12, 2014 to October 14, 2017, and data from October 15 to December 24, 2017 was selected as the test set to evaluate predictive effect. Results The coefficient of determination (R2) of the ARIMA (2,0,0) model is 0.87. The predicted values accord well with actual values.Conclusion The model analysis of ARIMA (2,0,0) shows that it has a good early warning effect in influenza surveillance in Chaoyang Hospital, Capital Medical University. This model can provide reference for other medical institutions in monitoring ILI cases.

Key words: influenza-like illness cases, monitor, autoregressive integrated moving average model

中图分类号: