The present study analyses the effect of a beverage composed of citrus and maqui (Aristotelia chilensis) with different sweeteners on male and female consumers. Beverages were designed and tested (140 volunteers) as a source of polyphenols, in a previous work. Plasma samples were taken before and after two months of daily intake. Samples were measured for bioactive-compound levels with metabolomics techniques, and the resulting data were analysed with advanced versions of ANOVA and clustering analysis, to describe the effects of sex and sweetener factors on bioactive compounds. To improve the results, machine learning techniques were applied to perform feature selection and data imputation. The results reflect a series of compounds which are more regulated for men, such as caffeic acid or 3,4-dihydroxyphenylacetic acid, and for women, trans ferulic acid (TFA) or naringenin glucuronide. Regulations are also observed with sweeteners, such as TFA with stevia in women, or vanillic acid with sucrose in men. These results show that there is a differential regulation of these two families of polyphenols by sex, and that this is influenced by sweeteners.
Keywords: anthocyanins; biostatistics; clustering; data imputation; dietary intake; feature selection; flavanones; machine learning.