DESENHO DE MODELOS DE PREDIÇÃO DA ATIVIDADE ANTIOXIDANTE DE MEL DE ABELHAS SEM FERRÃO

Abstract

The honey produced by native or stingless bees is used in healthy food and in traditional medicine. These honeys have a mildly sweet flavor with antiseptic, anti-inflammatory and healing properties in the treatment of infected wounds, in addition to having other properties such as antimicrobial, antiviral, anticancer and antioxidant. In Brazil, we can find more than 192 species of native bees. These bees, compared to Apis bees, are morphologically of medium or small sizes, therefore, their honey production is reduced. Its attractive nutraceutical properties and its low production make these honeys have a higher cost compared to honey produced by Apis bees. These facts contribute to its adulteration and currently there is not a proper regulation and standardization legislation for this type of honey. The main purpose of this work was to build statistical models to estimate the antioxidant capacity and proline content in 94 samples of stingless bee honey from 4 Brazilian states. Thus, the antioxidant capacity was first determined by the method of reduction of the stable radical DPPH (2,2-diphenyl 1-picrylhydrazine) and reduction of Fe3+ to Fe2+ (FRAP) and total amino acid content in proline equivalent. Next, digital images of the honeys were obtained, which were converted into RGB format data and then built the correlation models using regression by partial least squares (PLS). The models generated by PLS showed parameters of merit, with correlation coefficient, RMSEC, RMSECV, RMSEP and RMSEP/RMSECV ratio, suitable for the determination of antioxidant activity and proline content of new honey samples. Thus, the multivariate calibration of digital honey photography data is a viable tool to build prediction models that can be used to indirectly predict the antioxidant activity content and proline content in honey samples from stingless bees.