2005—2023年南昌市丙型肝炎流行特征及基于贝叶斯结构时间序列模型的发病趋势预测  

Epidemiological characteristics of hepatitis C and incidence trend prediction based on the BSTS model,Nanchang,2005-2023

作　　者：莫佳丽 涂志斌[3] 肖勇 路亮[3] 刘星雨 况杰[1,2] MO Jia-li;TU Zhi-bin;XIAO Yong;LU Liang;LIU Xing-yu;KUANG Jie(Department of Epidemiology,School of Public Health,Nanchang University,Nanchang,Jiangxi 330006,China;不详)

机构地区：[1]南昌大学公共卫生学院流行病学教研室,江西南昌330006 [2]疾病预防与公共卫生江西省重点实验室 [3]南昌市疾病预防控制中心 [4]南昌市青云谱区疾病预防控制中心

出　　处：《现代预防医学》2025年第4期583-589,共7页Modern Preventive Medicine

基　　金：国家自然科学基金(82160645,82360667);江西省自然科学基金(20212BAB206091);江西省卫生健康委科技计划(202410613)。

摘　　要：目的分析2005—2023年南昌市丙型肝炎流行特征,并探讨贝叶斯结构时间序列(BSTS)模型预测丙肝发病趋势的应用价值,为该市丙肝防控提供科学依据。方法收集2005年1月至2023年12月南昌市丙肝发病数据,采用时间序列分解法解析月发病数据的趋势和季节组分。运用R软件构建BSTS模型,其中2005年1月至2022年12月的数据作为训练集拟合BSTS模型,2023年1—12月数据作为测试集评估模型的预测效果,并将其预测准确性与自回归整合移动平均(ARIMA)模型进行比较,采用平均绝对误差(MAE)、平均绝对百分比误差(MAPE)、均方根误差(RMSE)和均方根百分比误差(RMSPE)评估预测准确性。结果南昌市丙肝发病总体呈上升趋势,2017年后发病数的增加有所减缓。发病数显示出明显的周期性和季节性变化,发病高发于春季3—5月,1—2月为低谷。BSTS模型预测性能指标MAE=9.67、MAPE=17.61%、RMSE=11.99和RMSPE=17.84均小于ARIMA模型预测性能指标MAE=12.12、MAPE=29.03%、RMSE=15.30、RMSPE=33.62。基于BSTS模型预测的2024年1月至2024年12月南昌市丙肝发病总数为308(95%CI:68~583)例,月均发病数为26(95%CI:6~48)例。结论南昌市丙肝发病存在周期性和季节性波动;BSTS模型预测性能更优,BSTS模型可为丙肝精准防控提供技术支撑。Objective To analyze the epidemiological characteristics of hepatitis C in Nanchang from 2005 to 2023 and evaluate the application value of the Bayesian structural time series(BSTS)model in predicting hepatitis C incidence trends,providing a scientific basis for the prevention and control of hepatitis C in the city.Methods Data on hepatitis C incidence in Nanchang from January 2005 to December 2023 were collected.The trend and seasonal components of the monthly incidence data were analyzed using time series decomposition.The BSTS model was constructed using R software,with data from January 2005 to December 2022 used as the training set to fit the model,and data from January to December 2023 used as the test set to evaluate the model's predictive performance.The prediction accuracy of the BSTS model was compared with that of the autoregressive integrated moving average(ARIMA)model using mean absolute error(MAE),mean absolute percentage error(MAPE),root mean square error(RMSE),and root mean square percentage error(RMSPE)as evaluation metrics.Results The overall incidence of hepatitis C in Nanchang has been on an upward trend,though the rate of increase has slowed since 2017.The data reveals significant cyclical and seasonal fluctuations,with a peak incidence in the spring months from March to May and a trough from January to February.The prediction performance indicators of the BSTS model(MAE=9.67,MAPE=17.61%,RMSE=11.99 and RMSPE=17.84)were all lower than those of the ARIMA model(MAE=12.12,MAPE=29.03%,RMSE=15.30,RMSPE=33.62).Based on the BSTS model,the total predicted number of hepatitis C cases in Nanchang from January 2024 to December 2024 is 308(95%CI:68-583),with an average monthly incidence of 26 cases(95%CI:6-48).Conclusion Hepatitis C incidence in Nanchang exhibits periodic and seasonal fluctuations.The BSTS model outperforms the ARIMA model in prediction performance and can provide technical support for the precise prevention and control of hepatitis C.

关 键 词：丙型肝炎 发病 预测 贝叶斯结构时间序列模型 自回归整合移动平均模型 

分 类 号：R183.9[医药卫生—流行病学]