基于运气理论研究北京地区流行性脑脊髓膜炎发病与气象因素的相关性

目的依据中医五运六气学说，研究北京地区流行性脑脊髓膜炎（以下简称＂流脑＂）的高发与同期及前期（包括1~3年前）气象因素的关联性，并建立BP人工神经网络的医疗气象预测模型。方法基于北京地区1970年~2004年35年的气象资料和流脑发病资料，利用BP人工神经网络方法，从不同时间维度分别建立流脑的气象预警模型。结果流脑的高发期主要集中在每年的初之气（即2、3月份）。利用当年、1年前、2年前、3年前的气象因素皆可成功建立流脑高发的预测模型，其预测精度分别是50%、75%、86%、99%,其中贡献度最大的气象因素分别是当年初之气的平均相对湿度、1年前五之气的平均气温、2年前初之气的平均风速、3年前五之气的平均风速。结论北京地区流脑的高发与当年及前3年的气象因素均具有关联性，利用前期（尤其是3年前）的气象因素建立传染病高发的预测模型具有深入研究的价值。

Objective To investigate the correlation between high incidence of epidemic cerebrospinal meningitis (ECM) and meteorological factors during the same and previous periods based on the theory of five movements and six climatic changes, and then to establish the medical meteorological predictive models of BP artificial neural network. Methods BP artificial neural network analysis was adopted to establish the ECM predictive models from different temporal dimensions, based on the data of meteorological factors and ECM from 1970 to 2004. Results The highest incidence of ECM was in the 1stqi (i.e. February and March). All predictive models can be successfully established with meteorological data of the year, one year before, two years before and three years before the ECM with the predictive accuracy of50%,75%,86% and 99% respectively. The study also shows the most significant meteorological factors include the average relative humidity in the 1stqi of the year, the average temperature in the 5thqi one year prior to the ECM, the average wind speed in the 1stqi two years prior to the ECM, and the average wind speed in the 5thqi three years prior to the ECM. Conclusion The high incidence of ECM in Beijing is related to meteorological variables in current year and the past three years. Furthermore, the predictive models with meteorological variables of previous years (especially three years prior to ECM) are of value to be further investigated.