| 基于贝叶斯时空模型的猩红热发病影响因素研究 |
| |
| 引用本文： | 李芳,宋秋月,陈佳,张彦琦,刘岭,易东,伍亚舟.基于贝叶斯时空模型的猩红热发病影响因素研究[J].现代预防医学,2022,0(6):963-968. |
| |
| 作者姓名： | 李芳 宋秋月 陈佳 张彦琦 刘岭 易东 伍亚舟 |
| |
| 作者单位： | 陆军军医大学军事预防医学系军队卫生统计学教研室，重庆400038 |
| |
| 摘 要： | 目的 猩红热（scarlet fever）是我国重点防制的法定乙类传染病之一，严重危害人类健康，其发病数据表现出典型的时空特征，利用时空分析方法研究猩红热发病的影响因素，为疾病防治工作提供帮助。方法 收集2014—2017年我国31个省市猩红热月发病资料及相应的气象因素和空气污染数据，利用Spearman相关分析和逐步回归分析筛选影响猩红热发病的气象因素和空气污染变量，将贝叶斯时空模型应用于猩红热发病的影响因素分析。结果 CO和NO2是对猩红热发病影响较大的因素，每升高一个单位分别会带来相对风险5%（95%CI:3.2%～6.91%）和1.66%（95%CI:1.05%～2.22%）的增加。比较无时空效应的一般贝叶斯模型的应用效果，贝叶斯时空模型优势明显，在训练集上，DIC = 9 205.526，MAE = 1.06，RMSE = 1.43，在测试集上，DIC = 3 060.089，MAE = 0.92，RMSE = 1.21，皆表现出优异的拟合能力和预测性能。结论 贝叶斯时空模型可有效研究疾病与相关因素之间的关系，在公共卫生领域对疾病预测及分析决策提供科学支撑和理论依据。
|
| 关 键 词： | 贝叶斯时空模型 猩红热 气象因素 空气污染 |
Influencing factors of scarlet fever based on Bayesian spatio-temporal model |
| |
| LI Fang,SONG Qiu-yue,CHEN Jia,ZHANG Yan-qi,LIU Ling,YI Dong,WU Ya-zhou.Influencing factors of scarlet fever based on Bayesian spatio-temporal model[J].Modern Preventive Medicine,2022,0(6):963-968. |
| |
| Authors: | LI Fang SONG Qiu-yue CHEN Jia ZHANG Yan-qi LIU Ling YI Dong WU Ya-zhou |
| |
| Affiliation: | Department of Army Health Statistics, Department of Military Preventive Medicine, Army Medical University, Chongqing 400038, China |
| |
| Abstract: | Objective Scarlet fever is one of the statutory category B infectious diseases in China and seriously endangers human health, with typical spatio-temporal characteristics. The spatio-temporal analysis method is used to investigate the influencing factors of scarlet fever incidence so as to provide help for disease prevention and control. Methods Monthly incidence data of scarlet fever and corresponding meteorological factors and air pollution data were collected from 2014 to 2017 in 31 provinces and cities in China, and meteorological factors and air pollution variables affecting the incidence of scarlet fever were screened using Spearman correlation analysis and stepwise regression analysis, and Bayesian spatio-temporal models were applied to the analysis of the influencing factors of scarlet fever incidence. Results CO and NO2 were the factors that had a greater effect on the onset of scarlet fever, with each unit increase bringing a relative risk increase of 5% (95%CI: 3.2%-6.91%) and 1.66% (95%CI: 1.05%-2.22%), respectively. Comparing the application effects of the general Bayesian model without temporal effects, the Bayesian temporal model had obvious advantage, with DIC=9 205.526, MAE=1.06, and RMSE=1.43 for the training set, and DIC=3 060.089, MAE=0.92, and RMSE=1.21 for the test set, all showing excellent fitting and prediction performance. Conclusion Bayesian spatio-temporal models can effectively study the relationship between diseases and related factors, and provide scientific support and theoretical basis for disease prediction and analysis decisions in the field of public health. |
| |
| Keywords: | Bayesian spatio-temporal model Scarlet fever Meteorological factors Air pollution |
|
| 点击此处可从《现代预防医学》浏览原始摘要信息 |
|
点击此处可从《现代预防医学》下载全文 |
|
