基于ARIMA模型预测性传播HIV/AIDS新发现病例数变化趋势  

Forecast of the trend of sexually transmitted HIV/AIDS cases based on ARIMA model

作　　者：尹诗琪 闫翔宇 苏鹤轩 张波 YIN Shiqi;YAN Xiangyu;SU Hexuan;ZHANG Bo(Shanghai Pudong New Area Center for Disease Control and Prevention,Pudong Institute of Preventive Medicine,Fudan University,Shanghai 200136,China;School of Public Health,Peking University,Beijing 100191;Medical Informatics Center,Peking University,Beijing 100191)

机构地区：[1]上海市浦东新区疾病预防控制中心复旦大学浦东预防医学研究院,上海200136 [2]北京大学公共卫生学院,北京100191 [3]北京大学医学信息学中心,北京100191

出　　处：《中国艾滋病性病》2020年第7期709-713,共5页Chinese Journal of Aids & STD

基　　金：上海市浦东新区卫生系统重点专科建设资助(PWZzk 2017-13)。

摘　　要：目的建立异性性传播和同性性传播艾滋病病毒(HIV)感染者/艾滋病(AIDS)(简称HIV/AIDS)的新发现病例数ARIMA时间序列模型,探讨两类性传播途径感染人群的新发现病例数变化趋势预测。方法利用中国疾病预防控制中心报告的全国2013年1月至2017年12月异性性传播和同性性传播途径HIV/AIDS新发现病例数资料建立ARIMA模型,对2018年1月至2018年9月不同性传播途径HIV/AIDS新发现病例数进行预测,对已建立的模型进行评价。结果异性性传播和同性性传播序列均呈现以年为周期(12个月)的波动上升。ARIMA(0,0,0)×(0,1,0)12模型更符合异性性传播规律,模型为:(1-B12)Yt=0.100+εt;ARIMA(0,1,1)×(0,1,0)12模型更符合同性性传播的规律,模型为:(1-B)(1-B12)Yt=(1-0.759B)εt。模型验证结果显示,预测结果基本符合实际新发现病例数的变化趋势,异性和同性性传播序列模型平均绝对误差(MAE)分别为830.28和374.10,平均相对误差(MAPE)分别为9.85%和15.52%。结论ARIMA模型可用于预测性传播HIV/AIDS新发现病例数。在后期研究中可进一步探讨新发现病例数在每周期波动的成因,并进一步优化参数,提高预测精准度。Objective To establish an ARIMA time series model of new cases of heterosexual and homosexual HIV/AIDS transmission,and explore the methods for predicting the trend of new cases in two types of sexual transmission routes.Methods The data of new cases of heterosexual and homosexual HIV/AIDS transmission from Jan 2013 to Dec 2017 reported by the Chinese Center for Disease Control and Prevention were used to establish ARIMA model.We predicted the number of new HIV/AIDS cases of the two routes from Jan 2018 to Sept 2018 to evaluate the established model.Results Both heterosexual and homosexual transmission sequences exhibited rising volatility in the annual cycle(12 months).The ARIMA(0,0,0)×(0,1,0)12 model was more consistent with the trend of heterosexual HIV/AIDS transmission cases and the model was:(1-B1)Yt=0.100+εt;ARIMA(0,1,1)×(0,1,0)12 model was more consistent with the trend of homosexual transmission and the model was:(1-B)(1-B1)Yt=(1-0.759 B)εt.The model verification results showed that the prediction results were basically consistent with the trend of the actual number of new cases.The Mean Absolute Error(MAE)of the heterosexual and homosexual transmission sequence models were 830.28 and 723.00 respectively;and the Mean Absolute Percentage Error(MAPE)were 9.85%and 15.52%respectively.Conclusion The ARIMA model can be used to predict the number of new cases of sexually transmitted HIV/AIDS.In the future studies,we can further explore the causes of fluctuations in the number of new cases,and further optimize the parameters to improve the accuracy of prediction.

关 键 词：艾滋病 性传播途径 新发现病例数 时间序列分析 自回归滑动平均求和模型 

分 类 号：R512.91[医药卫生—内科学] R373.9[医药卫生—临床医学]