基于ARIMA模型的地区武装冲突期间埃博拉病毒病疫情发展趋势预测  

Prediction of epidemic trends of Ebola virus during regional armed conflicts based on the ARIMA model

作　　者：陆诗阳 李银红 田玉洁 陈丹田 王磊 LU Shi-yang;LI Yin-hong;TIAN Yu-jie;CHEN Dan-tian;WANG Lei(Institute of Health Service and Transfusion Medicine,Academy of Military Medical Sciences,Academy of Military Sciences,Beijing 100850,China;Beijing University of Chemical Technology,Beijing 100029,China;Medical School of Chinese PLA,Beijing 100853,China)

机构地区：[1]军事科学院军事医学研究院卫生勤务与血液研究所,北京100850 [2]北京化工大学,北京100029 [3]解放军医学院,北京100853

出　　处：《军事医学》2023年第6期417-421,426,共6页Military Medical Sciences

摘　　要：目的探讨差分自回归移动平均(ARIMA)模型预测刚果(金)埃博拉病毒病疫情的可行性,分析武装冲突对疫情传播的影响,为防控工作提供参考依据。方法收集刚果(金)2019年9月至2020年2月埃博拉病毒病累计新增确诊病例数据,利用拟合函数求导处理,转换为每日新增确诊病例数并建立ARIMA模型,预测确诊数和死亡数并对模型预测的准确性进行评估。结果在平台期内北基伍省和伊图里省每日新增确诊病例的最优ARIMA模型分别为ARIMA(3,2,2)和ARIMA(1,2,6),预测的数据和拟合值吻合;进入冲突期后,因ARIMA预测值未考虑冲突影响,两地每日新增确诊病例数高于考虑了冲突的拟合结果。结论在不考虑武装冲突的情况下,ARIMA模型预测刚果(金)埃博拉病毒病疫情拟合效果相对较好,武装冲突可致每日新增确诊病例降低。该模型在埃博拉病毒病的预测预警中起到辅助作用,可为疫情的控制和医疗资源的合理调配提供参考。Objective To explore the feasibility of the ARIMA model being used to predict the Ebola hemorrhagic fever epidemic in Democratic Republic of the Congo(DRC),and to assess the impact of armed conflicts on the spread of the epidemic so as to provide reference for disease prevention and control.Methods The data related to Ebola haemor⁃rhagic fever in DRC was collected between September 2019 and February 2020.The derivative processing of fitting function was used to convert the data on daily newly confirmed cases and establish an ARIMA model to predict the number of confirmed cases and deaths and evaluate the accuracy of ARIMA model prediction.Results During the peak of cases,the best ARIMA model for the daily newly confirmed cases in North Kivu and Ituri provinces was ARIMA(3,2,2)and ARIMA(1,2,6),for the predicted data was consistent with the fitting value.However,during armed conflicts,the predicted value by ARIMA,which excluded the impact of conflicts,was higher than the fitting result that took conflicts into consideration.Conclusion The fitting effect of the ARIMA model is better when used to predict the epidemic trends of the Ebola haemorrhagic fever in DRC if armed conflicts are not taken into account.Armed conflicts can reduce the number of daily newly confirmed cases.This model can assist the prediction and early warning of Ebola haemorrhagic fever,and provide reference for epidemic control and proper allocation of health care resources.

关 键 词：ARIMA模型 埃博拉 预测 刚果(金) 

分 类 号：R373.32[医药卫生—病原生物学]