2005—2021年佳木斯市流行性腮腺炎流行病学特征与预测模型研究  

Study on Epidemiological Characteristics and Prediction Model of Mumps in Jiamusi from 2005 to 2021

作　　者：王小亭 李明春[2] 周宪君[1] Wang Xiaoting;Li Mingchun(School of Public Health,Jiamusi University,Jiamusi,Heilongjiang,154007,China)

机构地区：[1]佳木斯大学公共卫生学院,黑龙江佳木斯154007 [2]佳木斯市疾病预防控制中心,黑龙江佳木斯154007

出　　处：《中国初级卫生保健》2024年第1期91-95,共5页Chinese Primary Health Care

摘　　要：目的:分析佳木斯市流行性腮腺炎流行病学特征,探讨差分自回归移动平均(ARIMA)模型和随机森林模型在佳木斯市流行性腮腺炎流行趋势预测中的应用。方法:通过中国传染病报告信息管理系统收集2005—2021年佳木斯市流行性腮腺炎报告数据,对其进行描述性流行病学分析。选取2005—2018年佳木斯市流行性腮腺炎月发病数据构建ARIMA模型和随机森林模型进行训练和预测,比较两种模型的效果评价指标。结果:2005—2021年佳木斯市共报告流行性腮腺炎3020例,年均报告发病率为7.34/10万,发病呈季节性双峰分布,主高峰为11月—次年1月,次高峰为4—7月;流行性腮腺炎病例主要集中在0~20岁组,共报告流行性腮腺炎病例2777例,占全部病例数的91.95%;发病居前3位的人群分别是学生、幼托儿童和散居儿童,共报告2741例,占全部病例数的90.76%。最优模型ARIMA(4,1,3)(1,1,1)12和随机森林模型预测的均方根误差(RMSE)和相对误差分别是16.65、14.69和38.78%、34.50%。结论:中小学生和散居儿童是流行性腮腺炎防控的重点人群,应做好学校卫生管理工作,加强学校流行性腮腺炎疫情监测。ARIMA(4,1,3)(1,1,1)12模型是预测佳木斯市流行性腮腺炎月发病趋势较为理想的模型。OBJECTIVE To analyze the epidemiological characteristics of mumps in Jiamusi.To explore the applications of au⁃toregressive integrated moving average model(ARIMA)and Random Forest(RF)models for the prediction of the prevalence trend of facial diseases in Jiamusi.METHODS Descriptive epidemiological analysis was carried out by collecting the data of cheek reports from 2005 to 2021 in Jiamusi through China Infectious Disease Reporting Information Management System.The ARIMA model and the random forest model were constructed based on the number of mumps from 2005 to 2018 for training and prediction,and the evaluation indexes of the two models were compared.RESULTS From 2005 to 2021,a total of 3020 cases of gills were reported in Jimusi,with an average annual incidence of 5.43/105.The incidence presented seasonal bimodal distribution,with the main peak from November to January and the secondary peak from April to July.The cases were mainly concentrated in 0~20 years old group(91.95%).The top three groups were students,children in nursery care and scattered children(2741 cases).The root mean square errors(RMSE)and relative errors predicted by random forest and optimal model ARIMA(4,1,3)(1,1,1)12 were 16.65,14.69 and 38.78%,34.50%,respectively.CONCLUSION Primary and secondary school students and scattered children are the key population for the prevention and control of mills.School health management should be done well,and surveillance of mills should be strength⁃ened.ARIMA(4,1,3)(1,1,1)12 model is an ideal model for predicting the incidence of cheek moon in Jiamusi.

关 键 词：流行性腮腺炎 流行特征 差分自回归移动平均模型 随机森林 预测 

分 类 号：R181[医药卫生—流行病学]