考虑空间分层异质性的中国手足口病复杂时空风险贝叶斯建模研究

Hand foot and mouth disease has frequently occurred in China in recent years and its space time propagation process is complex, thus we should consider the influence of spatial stratified heterogeneity. While the traditional spatio-temporal analysis methods are simple and cannot meet the needs of complex spatio-temporal risk modeling of infectious diseases. Based on our previous researches and through the innovation driven of the improved spatio-temporal regression model, this project builds spatio-temporal modeling system considering spatial stratified heterogeneity under the Hierarchical Bayesian framework to deeply explore the complex space time risk differentiation of hand foot and mouth disease in china. The project is carried out in three progressive aspects including risk pattern, risk factors and risk prediction at spatio-temporal scale. Firstly, a geographical zoning spatio-temporal interaction model is proposed to mine the space time interaction disease risk considering spatial stratified heterogeneity. Further, a local spatio-temporal ecological regression model is built to identify multi-source disease risk factors and to fit the stratified heterogeneity and the local space time relative risk between factors and disease. At last, a spatio-temporal joint analysis model is established to predict the complete spatial and temporal shared disease risk distribution using both disease report and internet data sources under the spatial stratified heterogeneity condition. Through the basic research and development of the traditional spatio-temporal statistical models, this study purposes new methods for the analysis of spatial and temporal risk of infectious diseases, reveals the space time propagation of hand foot and mouth disease risk. It is of great significance to the prevention and control of hand foot mouth disease in China and has important theoretical value to the geographical epidemiology and spatio-temporal statistics.

手足口病近年来在我国流行严重且时空传播过程复杂，应考虑空间分层异质性的影响，但传统时空分析方法简单无法满足传染病复杂时空风险建模的需求。项目在前期研究基础上，在层次贝叶斯框架下搭建考虑空间分层异质性的时空建模体系，通过改进传统时空回归模型的创新驱动深入挖掘我国手足口病时空风险分异规律。项目具体从我国手足口病的时空风险格局、时空风险因子和时空风险预测三个角度开展递进式研究：首先建立地理分区时空交互模型，模拟考虑多尺度地理行政区划的时空多模态风险格局；然后建立局域时空生态回归模型，识别多源时空风险因子并挖掘因子的局域时空相对风险；最后建立时空联合分析模型，预测综合考虑疾病监测和网络来源的完整时空共享风险分布。通过发展传统时空统计模型的基础研究，形成传染病时空风险分析的新方法，揭示手足口病时空风险传播规律，对我国手足口病防控具有重要应用意义，对地理流行病和时空统计学具有重要理论价值。

儿童手足口传染病在中国流行严重且时空传播过程复杂，特别是在县域地理空间尺度。项目通过在层次贝叶斯框架下构建考虑空间分层异质性的空间、时空回归建模体系，深入挖掘我国县域手足口病的复杂时空风险分异规律。首先，项目创新提出渐进时空预测（PST）法，成功预测了中国首套完整的县市社会经济时空数据集，其指标作为后续手足口病建模的关键解释因素。然后，项目针对我国手足口传染病的地理空间分异格局、风险因子和风险预测三个角度逐步开展研究。具体来说，（1）创新提出考虑空间分层异质性的疾病相对风险降尺度（DRRD）模型，模拟了中国县域手足口病发生概率的相对风险地图集，评价指标包括空间相对危险度, 空间比值比和空间归因危险度，用于提示隐藏的区域额外风险，探测地理空间不公平性集聚区。（2）提出局域时空回归新方法—贝叶斯时空变系数（STVC）模型，成功挖掘六种典型气象风险因素在局域空间单元（县）和时间点（月）对手足口病发生风险的时空异质性影响效应。STVC模型是贝叶斯统计体系下首个适用于地理空间大数据计算的局域时空回归方法，能够有效处理并探测时空非平稳性，是进行局域时空影响因素分析和时空预测的新工具。（3）构建同时考虑气象和社会经济因素的时空零膨胀层次贝叶斯模型，拟合大面积零值地区的传染病潜在风险并填补缺失值，进而预测了中国县域手足口病发病率的完整时空风险地图集。进一步，利用广义加性回归模型，综合网络搜索引擎数据和气象环境因素，预测手足口病实时变化趋势，可作为传统疾病监测手段的补充。通过改进传统空间和时空回归统计模型的创新驱动，项目明确了手足口病在中国大陆地区的时空风险分布格局和相关风险因素，并预测其时空风险分布，对我国手足口疾病防控防治具有一定应用意义；同时，项目提出的前沿空间、时空统计新模型，对GIScience、时空统计和健康地理等交叉学科的发展具有重要基础科研价值。

内容获取失败，请点击重试