广州大都市登革热时空传播混合模式

Mixing spatial-temporal transmission patterns of metropolis dengue fever:A case study of Guangzhou, China

以广州市2014年登革热疫情早期11周内679个登革热病例数据为例,提出了一种基于扩展Knox检验的疾病时空传播网络模型,研究登革热疫情早期时空传播特征。首先,应用扩展Knox检验方法发现2周1 km内、1周5~7 km两个时空标度下,病例之间具有显著的时空交互特征;其次,提取两个时空标度下的病例点对,分别构造时空近邻传播（STAT）网络和日常移动传播（HDMT）网络;最后,对STAT和HDMT传播网络的同配性、空间特征、网络中心位置进行对比分析。结果显示,广州市登革热传播是空间近邻交叉传播和空间扩散传播的混合传播模式;利用STAT网络亲近中心度标识出的＂爆发中心＂与病例空间核密度中心基本重合,HDMT网络中介中心度标识出的＂扩散中枢＂多位于城市快速交通干线附近。

This paper proposes a new method to model the spatial and temporal transmission network for infectious disease. Specifically, 679 cases from the early 11 weeks of the dengue fever outbreak in Guangzhou in 2014 are used to analyze the disease transmission characteristics. Three methods are adopted for the analysis.（1） We use extended Knox test to derive the main time and space interaction sectors at a distance of 1 km in two weeks and that of 5-7 km in one week.（2） We pair the cases from different areas to construct the space-time affinity transmission（STAT） network and the human daily movement transmission（HDMT）network.（3） We compare the assortativity, spatial characteristics, and the central network location between STAT and HDMT network by using complex network theories. The result shows that the percentages of overall cases included in the STAT and HDMT networks are92.93% and 97.05%, respectively. This means that both STAT and HDMT network models imply the overall transmission of the dengue fever outbreak. STAT network is assortative, and presents the cross infection in neighboring areas. On the contrary, HDMT network is disassortative, and it displays the diffusion infection character of the dengue fever outbreak. We earmark the location of outbreak center as well as the diffusion center with the degree of closeness centrality in STAT network and the degree of betweenness centrality in HDMT network. This shows that the outbreak center approximately overlaps the spatial kernel density center of all cases, while the diffusion centers are located along the urban rapid transit routes.