基于人工神经网络的中国人心电图QT间期参考值的地理分布

摘要

目的：为制定中国人心电图 QT 间期参考值的统一标准提供依据。方法收集中国84个市（县）级医院、研究单位及高等院校测定的15164例健康成年人心电图 QT 间期参考值。运用相关分析、主成分分析、人工神经网络、GIS 空间分析等方法研究其与经度、纬度、海拔、年平均气温、年平均相对湿度、年降水量、气温年较差、表土粉粒百分率、表土阳离子交换量（黏土）、表土阳离子交换量（潜育土）、表土碱度的关系。结果通过五层人工神经网络模拟健康中国人心电图 QT 间期的参考值范围，并通过克里格插值法精确地内插出中国人心电图 QT 间期参考值范围的地理分布图。结论知道中国某地的地理要素，就可用此模型估算该地区成年人心电图 QT 间期参考值，同时也可用空间分布图直观得到中国任何地方的成年人心电图 QT 间期参考值。%Objective To supply a scientific basis for unifying the reference value standard of QT interval of Chinese people .Meth-ods The QT interval reference values were collected from 15 164 healthy people in 84 areas of China.The correlations between the measured values of QT interval and the 11 geographical factors were analyzed .Results After 5 layers′Neural Networks and self-study about building the analogue rules for 60 times, the relationship between the normal reference value of QT interval and geographic factors had be limitated with the methods of BP neural networks and Spatial analysis of GIS .The geography distribution map of the reference range of QT interval was inserted exactly with the method of Kriging .Conclusions If geographical factors of a particular area are obtained ,the QT interval reference value of Chinese people in this area could be calculated using this model and the normal QT interval reference values of Chinese men any -where in China could be obtained from geography distribution figure .