Estudo do perfil de contaminação de alimentos por dioxinas quanto à possível origem utilizando modelos quimiométricos discriminantes.

Abstract

Dioxin and furan contamination events are characterized by being complex scientific problems that require specific knowledge about production processes, survey of likely sources of contamination, understanding of bioaccumulation and transfer patterns, once the source of contamination has generally a specific fingerprint. Moreover, the contamination events demand a large amount of investment, time and labor effort to elucidate them. The need to mitigate, in the shortest time, an event creates the need to use tools that can classify the contamination profile, found in relation to a group of profiles which have a known origin. Profiles from various sources of contamination were compiled from the existing literature, being classified into twelve different groups. The discriminant analysis by partial least squares was used in order to build a model that would allow classifying profiles of samples of dioxins and furans as to the probable source of contamination. The developed model was evaluated for the following figures of merit: sensitivity, specificity, false positive rate and false negative rate. The model's sensitivity values ranged from 73.7 to 100%. Values for specificity ranged from 76.9 to 100%. False positive and false negative rates were up to 26.3 and 23.1%. Regression vector plots indicated which variables are the most significant for the classification of a given contamination profile for a given group. Profiles of poultry fat, eggs and feed samples, supplied by the Dioxins and PCB's Laboratory from the Federal Laboratory of Animal and Plant Health and Inspection, were classified, and the result obtained was consistent with the expected classification, although bioaccumulation and transfer rates are not considered. The developed model can be used to guide the investigation process, suggesting its probable origin, simplifying and speeding up the process of monitoring these food contaminants in Brazil.