中国痢疾发病率的时空分析及短期预测

目的  分析我国大陆31个省、市、自治区2004-2016年间痢疾发病率的时空相关性，预测全国痢疾短期发病率的效果。 方法  获取我国2004-2016年痢疾发病率资料。使用Arcgis10.5和Geoda软件（2018稳定版）制作可视化发病率分级地图并分析空间相关性，使用自回归移动平均（auto-regressive integrated moving average，ARIMA）模型预测2017年全国痢疾发病率并评价模型效果。 结果  我国2004-2016年痢疾发病率逐年降低，西部地区痢疾发病率普遍高于东部地区，但北京、天津发病率依然较高。发病率基本不存在全局相关，但存在局部聚集。青海由高高聚集转为低高聚集，内蒙古和山西由无局部聚集转为低高聚集，陕西长期呈高高聚集，东南沿海地区长期处于低低聚集。预测全国痢疾月发病率的模型为ARIMA（1，0，0）（2，1，1）₁₂模型，实际发病率均落在预测区间内。 结论  2004-2016年痢疾发病率在空间上没有明显的移动性但有聚集性，北京、天津、陕西及西部地区发病情况依然严峻。使用ARIMA模型可以很好的预测短期痢疾月发病率，应根据发病趋势和聚集情况以及短期预测结果综合制定防控措施。

Spatio-temporal analysis and short-term prediction of the incidence of dysentery in China

  Objective  The aim is to analyze the spatial-temporal correlation of dysentery incidence in 31 provinces, municipalities and autonomous regions in China from 2004 to 2016, and to predict the short-term incidence of dysentery in China.  Methods  Data about the incidence of dysentery from 2004 to 2016 was collected. Arcgis and Geoda were used to create visualized grading maps and analyze spatial correlation. The auto-regressive integrated moving average model (ARIMA)was used to predict the incidence of dysentery in 2017 and evaluate the prediction accuracy of the model.  Results  The incidence of dysentery in China declined with each passing year from 2004 to 2016. The incidence of dysentery in the western region was significantly higher than the eastern region, except high incidence rate in Beijing and Tianjin. There was no significantly global correlation in the incidence rate, but there was local aggregation. Qinghai had turned from high-level aggregation to low-level accumulation. Inner Mongolia and Shanxi had changed from no local aggregation to low-high accumulation. Shaanxi has long been high-high, and the southeast coastal areas had been low-low accumulation for a long time. The optimal model ARIMA (1, 0, 0) (2, 1, 1)₁₂ was established to predict the incidence of dysentery, and the prediction results were roughly consistent with the observations.  Conclusion  The incidence of dysentery from 2004 to 2016 is not spatially mobile but clustered. The incidence of dysentery in Beijing, Tianjin, Shaanxi and most of the western regions is severe. The ARIMA model is suitable for forecasting the incidence of short-term dysentery. And our analysis may help prevent and control the incidence of dysentery in China.