LAUSR

The objective of this study was to develop a method for authenticating organic milk samples in North Spain on the basis of its trace mineral composition. Fourteen elements in 98 samples were determined by inductively coupled plasma mass spectrometry. Although concentrations of Co, Cr, Cu, I, Se and Zn where statistically higher in conventional milk and As in organic, none of these elements by itself was able to discriminate between organic and conventional milk. The chemical data was examined by principal component analysis and cluster analysis, revealing a natural separation between organic and conventional milk. In a second step, several supervised pattern recognition techniques were used to construct mathematical models for predicting the type of milk (organic or conventional) based on the metal content. The results proved that the model constructed using the artificial neural network is capable of correctly identifying the type of milk in almost 95% of cases.