基于“三年化疫”理论探讨百日咳发病与前期气象因素的相关性并建立预测模型  

Exploration on the Correlation between the Incidence of Pertussis and Previous Meteorological Variables Based on the Theory of ″Plague Transformation in Three Years″ and Building Prediction Models

作　　者：张轩[1] 贺娟[1] 

机构地区：[1]北京中医药大学基础医学院,北京100029

出　　处：《西部中医药》2015年第11期38-42,共5页Western Journal of Traditional Chinese Medicine

基　　金：国家自然科学基金项目(编号81072896)

摘　　要：目的：探讨北京地区百日咳发病与前期（1~3年前）气象因素的关联性,并建立BP人工神经网络医疗气象预测模型。方法：收集北京地区1970—2004年35年的气象资料和百日咳发病资料,采用BP人工神经网络方法,分别从1年前、2年前、3年前3个不同时间维度,建立百日咳发病的气象预警模型。结果：1）百日咳发病与前期（1~3年前）气象因素具有相关性,其中关系最密切的气象因素是1~3年前的平均相对湿度;2）利用前期气象因素皆可成功建立百日咳发病的预测模型,以1年前气象因素建模的预测效果最佳。结论：1~3年前的气候可能会影响某些传染病的发病,今后对传染病发病的研究应注意考虑前期的气象变化。Objective： To explore the correlation between pertussis incidence in Beijing and previous （one year to three years ago） meteorological variables, and establish medical meteorological prediction models of BP artificial neural network. Methods： The data of climate change from 1970 to 2004 and pertussis incidence in Beijing were collected and meteorological prediction models were established by adopting the method of BP artificial neural network from three different time dimensions including one year, two years and three years ago respectively. Results： 1） Per- tussis incidence is related to previous （one year to three years ago） meteorological variables,, among them, closely related meteorological factor was average relative humidity （one year to three years ago）; 2） Prediction model ofpertussis incidence could be successfully established by using previous meteorological variables, especially prediction models established on the foundation of meteorological one year ago showing the best prediction effects. Conclusion： Climates one to three years ago might be affect the incidence of some infectious diseases, and the study on the incidences of infectious diseases should pay attention to meteorological changes early.

关 键 词：百日咳 BP人工神经网络 前期气象变化 三年化疫 五运六气 

分 类 号：R516.7[医药卫生—内科学]