基于半参数零膨胀时空回归模型研究流感季节性区域传播风险

Influenza is prevalent throughout the subtropics, but the mechanism of seasonal transmission in the region is unclear. The antigenic variation of influenza virus and meteorological environmental factors are considered to be the main factors affecting the seasonal and regional spread of influenza, but most of the previous studies have focused on meteorological and environmental factors, and less consideration has been given to the influence of antigenic variation of influenza virus. At present, the statistical modeling of influenza transmission risk is mainly based on parametric model, which has a strong subjectivity, and can not handle the problem of large deviation between model hypothesis and real data distribution. Combined with the "zero inflation" phenomenon of influenza activity surveillance data, this project proposes a semi-parametric zero-inflated spatiotemporal regression model, which relaxes the restrictions on specific distribution assumptions and obtains a more robust seasonal and regional influenza transmission risk estimation model. Taking Guangdong-Hong Kong-Macau Greater Bay Area as the research area, the degree of antigenic variation of influenza virus is quantified by the HA protein sequence of influenza virus. A robust semi-parametric zero-inflated spatiotemporal regression model was used to study the influence of antigenic variation of influenza virus on the risk of seasonal and regional transmission of influenza virus, which provides important evidence for further understanding the transmission of influenza virus in Guangdong-Hong Kong-Macau Greater Bay Area and subtropical regions.

流感在亚热带地区常年流行，然而在此区域内的季节性传播机制尚不明确。流感病毒抗原变异和气象环境因素被认为是影响流感季节性区域传播的主要因素，但已有研究多集中在气象环境因素，而较少考虑流感病毒抗原变异的影响。目前流感传播风险的统计建模以参数模型为主，无法解决模型假设与真实数据分布之间存在较大偏差的问题，具有很强的主观性。结合流感活动监测数据的“零膨胀”现象，本项目提出构建半参数零膨胀时空回归模型，放宽对特定分布假设的限制，得到更加稳健的流感季节性区域传播风险估计模型。本项目以粤港澳大湾区为研究区域，通过流感病毒HA蛋白序列量化流感病毒抗原变异程度，利用稳健的半参数零膨胀时空回归模型研究流感病毒抗原变异对流感季节性区域传播风险的影响，为进一步研究粤港澳大湾区乃至亚热带流感病毒传播情况提供重要证据。

在热带和亚热带地区，流感全年都有流行，气象环境因素无法单独解释流感季节性传播风险，因此，针对热带和亚热带地区研究流感病毒抗原变异对流感季节性区域传播风险影响有着重要意义，本项目提出构建零膨胀时间序列回归模型对流感活动监测数据进行分析。. 项目按计划圆满完成了预期的研究目标，主要研究内容包括：（1）明确粤港澳大湾区内流感病毒抗原变异情况，建立流感病毒株 HA 蛋白序列数据库；（2）计算出流感病毒抗原变异程度(AVI)；（3）零膨胀时间序列回归模型的构建；（4）参数估计；（5）模拟研究，进行模型验证；（6）开发零膨胀时间序列模型R包。. 本项目成果主要体现在论文发表，项目共发表SCI论文11篇，其中1篇被国际指南收录。此外尚有2篇SCI论文正在投稿当中。. 本项目的意义在于：探究了流感病毒抗原变异对流感季节性区域传播风险的影响，从而为流感等重大及新发突发传播病的预警提供新策略。.