北京地区肠道传染病与气象因素的关联性分析

Correlation analysis on gastrointestinal diseases and meteorological variables in Beijing

目的 依据中医五运六气学说,研究北京地区肠道传染病发病与当年及1~3年前气象因素的关联性,并尝试建立气象预警模型。方法 基于北京地区1970—2004年35年的气象数据和肠道传染病（包括细菌性痢疾、病毒性肝炎、伤寒副伤寒）发病数据,利用BP人工神经网络统计方法,建立＂气象—肠道传染病＂的预测模型,并筛选出影响发病的关键气象因子。结果 利用当年及1,2,3年前的气象因素可分别建立肠道传染病发病的气象预测模型,预测精度均约为60%,其中贡献度较大的气象因素分别是当年的相对湿度,以及1~3年前的相对湿度和风速。利用上述当年~3年前的关键气象因子建模,其预测效果最佳,精度约为80%。结论 北京地区肠道传染病发病不仅与当年的气候变化有关,还与1~3年前的气象变化具有相关性;利用当年~3年前（共4年）的气象因素建立预测模型,效果要优于单用某一年的气象因子所建模型。

Objective According Chinese medicine theory＂five circuits and six qi＂to study the correlation between the incidence of gastrointestinal diseases in Beijing area and meteorological factors in current year and the past three years based on theory of Yunqi,so as to establish meteorological forecasting model. Methods The back-propagation artificial neural net-work was conducted using data on five basic meteorological factors and data on incidence of gastrointestinal diseases（ including dysentery,typhoid fever and paratyphoid fever,viral hepatitis） in Beijing area,for the period 1970-2004. Results All predictive models using meteorological factors during the corresponding year and one to three years ago can be successfully established and the forecast accuracy were nearly 60%. Additionally,the meteorological factors with largest contribution include average relative humidity in the current year,and average relative humidity as well as average wind speed in one to three years ago. The model using these vital meteorological factors has shown a relatively high accuracy rate 80%.Conclusion The incidence of gastrointestinal diseases in Beijing is related to meteorological variables in current year and the past three years. Furthermore,the predictive models suggest that the use of a collection of meteorological variables spanning from three years ago to the present is superior to the use of data from only a single year using.