微量元素谱的计算机判别分析──食管癌高中低发区人尿元素谱的分类识别
Study of Computer Discriminant Analysis on Personal Urine TE SpectumClassfication of Personal Urine TE Spectrum in the High,Madial and Low Incidence of Esophgeal Cancinoma Districts
Abstract
应用计算机对汕头地区的食管癌高中低发区的204份健康人尿样的微量元素谱进行单因素的均值方差分析和Q检验.用90例的观测值建立判别函数,应用该函数识别不同地区的微量元素谱,其判别效应经方差分析,P值<0.01,回判准确率83.3%;对同时测定的114例观测值进行分类判别,识别的准确率达76.3%. The contents of 16 trace elements(TE) in 90 urinary samples collected from the high, medial and low incidence of esophageal cancinoma(EC) districts (Shantou area)were variance and multivariable stepwise discriminant analysed by computer.The stepwise discriminant equations were established by 8 variables.Applying the discriminant equation to the high,medial and low incidence.districts by resubstitution test 83.3% accuracy and to the other 114 samples accuracy is 76.3% Thus,these equations may provide a basis for prediction high,madial and low incidence of Esophageal cancer.