Knight G J, Heese H de V, Dempster W S, Kirsten G
Pediatr Res. 1983 Feb;17(2):168-70. doi: 10.1203/00006450-198302000-00017.
Hematologic variables were measured in 240 apparently healthy infants ranging from 1-12 months of age attending a well baby clinic. There were 20 infants for each month of age. Hematologic parameters were measured in each infant by Coulter Counter Model S. Serum iron, total iron binding capacity, free erythrocyte protoporphyrin (FEP) and serum ferritin levels were measured in most infants. Their weights together with their serum iron, total iron binding capacity, and serum ferritin were judged to be independent variables of iron status, whereas the hematologic variables were considered to be response variables indicative of iron status. The correlation coefficients among these variables, after excluding redundant variables and transforming to logarithms, were computed. Canonical correlation analysis was applied to the matrix of correlation coefficients to yield the linear function of the independent variables most highly correlated with a linear function of the response variables. The linear function of the response variables was found to be well approximated by the logarithm of the mean corpuscular haemoglobin, which was highly correlated with each of the independent variables.
在一家健康婴儿诊所,对240名年龄在1至12个月的看似健康的婴儿进行了血液学变量测量。每个月龄有20名婴儿。使用库尔特S型血细胞计数器对每个婴儿进行血液学参数测量。大多数婴儿测量了血清铁、总铁结合力、游离红细胞原卟啉(FEP)和血清铁蛋白水平。他们的体重以及血清铁、总铁结合力和血清铁蛋白被判定为铁状态的自变量,而血液学变量被视为指示铁状态的反应变量。在排除冗余变量并转换为对数后,计算这些变量之间的相关系数。对相关系数矩阵应用典型相关分析,以得出与反应变量的线性函数高度相关的自变量的线性函数。发现反应变量的线性函数可以很好地用平均红细胞血红蛋白的对数来近似,而平均红细胞血红蛋白的对数与每个自变量都高度相关。