This study successfully applied a potentiometric E-tongue with 20 cross-selectivity lipidic polymeric membranes in the discrimination of three semi-quantitative groups, that represented the following intervals of honey adulteration percentage with cane sugar: 0 %; [0, 10]%; [10, 20]% of adulteration. We analysed five different types of Portuguese honey; five brands of cane sugar were added to the adulterated samples; a comparative analysis was then performed. Linear discriminant analysis coupled with a tabu search algorithm for feature selection was applied to the ETongue's analytical data to select the best model. A discriminant model with 12 sensors was obtained. This model classified correctly all samples in both in internal (train data, 15 samples) and external validation (test data,10 samples). Also, multiple linear regression with tabu search was applied to verify if ETongue's data would allow quantifying the honey's adulteration level. The results showed that it was possible to obtain a quantitative model but with unsatisfactory predictive performance in the test data group (external validation), giving, in general, values below the expected concentrations. E-tongue is a real-time green, flexible and low-cost analytical tool that requires minimum sample preparation and no special technical skills, being a promising tool for everyday application.
This work was carried out in the course of ‘Electronic Tongue and Nose in Food Technology’ of the Master's degree in Food Technology of Universidade Tecnológica Federal do Paraná, Campo Mourão, Brasil. This study was financed in part by Conselho Nacional de Pesquisa e Desenvolvimento Tecnológico (CNPq, 308153/2018‐9) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior ‐ Brasil (CAPES) – Finance Code 001. The Portuguese authors are grateful to the Foundation for Science and Technology (FCT, Portugal) for the financial support through national funds FCT/MCTES (PIDDAC) to CIMO (UIDB/00690/2020 and UIDP/00690/2020) and SusTEC (LA/P/0007/2021).
