Factorial discriminant analysis of occupational exposure in metallurgy using INAA of hair samples

Abstract

A multivariate statistical technique-factoral discriminant analysis (FDA)-has been used to evaluate the occupational esposure of a group of metallurgy workers using the elemental composition of their scalp hair, as compared with a control group. The initial variables were the concentrations of ten minor and trace elements Al, Co, Cu, Fe, Mg, Mn, Sb, Se, V, and Zn determined in hair samples by using instrumental neutron activation analysis (INAA). A good discrimination between exposed and control subjects has been achieved after the normalization of the data (by log-transforming the initial variables) and considering sex-matched groups. To measure the accuracy of the classification and the predictive ability a cross-validation procedure was used.