**Background:** Metabolic diseases are linked to high-sugar beverage consumption, driving demand for plant-based alternatives with health-promoting properties. (Poly)phenols, found in fruits like maqui berry and citrus, have known antioxidant, anti-inflammatory, and cardiovascular benefits. However, the bioavailability of these compounds—how they are absorbed, metabolized, and excreted—can be influenced by formulation (e.g., sweetener type) and consumer characteristics (e.g., sex). Historically, nutritional trials have not considered sex as a relevant factor, potentially biasing results. This study extends prior work on plasma samples to urine samples, aiming to elucidate differential metabolism of flavonoids between sexes using advanced biostatistical and machine learning methods.
**Methods:** A two-month longitudinal intervention trial was conducted with 138 overweight volunteers who consumed a maqui-citrus beverage daily. The beverage was sweetened with one of three sweeteners: sucrose, sucralose, or stevia. Urine samples were collected at baseline (day 0) and at the end of the intake period (day 60). Phenolic metabolites were identified and quantified using UHPLC-ESI-QqQ-MS/MS. The metabolites were classified into two sets: (poly)phenol metabolites (including caffeic acid, 3,4-dihydroxyphenylacetic acid (DHPAA), ferulic acid, vanillic acid, and their derivatives) and flavanone metabolites (including eriodictyol, homoeriodictyol, naringenin, and their phase II metabolites). The computational analysis used R Statistical Software v4.1.1 and included: data preprocessing and normalization, three-way paired ANOVA (factors: time, sex, sweetener) followed by multiple pairwise t-tests with Benjamini–Hochberg correction, multivariate data imputation, feature selection using Boruta's algorithm, and clustering analysis using the PAM algorithm with the number of clusters (k=6) selected by NbClust and clValid packages.
**Key Results:** For (poly)phenol metabolites, ANOVA revealed a significant effect of time (beverage consumption) on DHPAA (p=0.002), VA (p=0.025), VA-GS (p=0.028), and Total VA (p=0.047). DHPAA showed the highest increase in concentration over time, while VA-GS decreased. The sex–time interaction was significant for CA (p=0.024) and DHPAA (p=0.0043), with women showing greater increases than men; DHPAA actually decreased in men. The sweetener–time interaction was significant for DHPAA (p=0.0026) and DHPAA-GS (p=0.03). Pairwise t-tests identified a strong positive effect of stevia on DHPAA in men (p<0.001) and a weaker positive effect in women (p<0.05). For flavanone metabolites, ANOVA showed a significant time effect on HE (p=0.013), HE-G (p=0.001), HE-GG (p=0.008), NG (p=0.003), Total HE (p=0.00066), and Total N (p=0.003). All increased over time, with HE-G and Total HE showing the largest increases. The sex–time interaction was significant only for ES (p=0.012), with concentrations increasing in women. The sweetener–time interaction was significant for E (p=0.027) and ES (p=0.028). Pairwise t-tests revealed that stevia positively influenced ES, HE-G, and Total HE in women, and NG and Total N in men. Clustering analysis of (poly)phenol metabolites at baseline identified two main clusters: cluster 3 (48 individuals, balanced sexes, low metabolite levels) and cluster 4 (balanced sexes, high Total CA). At the end of the trial, clusters with more women and alternative sweeteners showed enhanced (poly)phenol metabolite bioavailability, including high Total CA and CA metabolites (DHPAA, TFA). Clusters with more men showed high Total CA but low metabolites. For flavanone metabolites at baseline, cluster 1 (mostly women) had high naringenin derivatives and low eriodictyol/homoeriodictyol derivatives, while clusters 3 and 5 (mostly men) had high eriodictyol and homoeriodictyol. At the end, sucrose appeared to enhance flavanone bioavailability overall, with stronger effects in women.
**Clinical Implications:** This study demonstrates that both the consumer's sex and the type of sweetener used in (poly)phenol-rich beverages significantly affect the bioavailability of health-promoting metabolites. Stevia emerged as a promising alternative to sucrose, enhancing the bioavailability of several key metabolites, including DHPAA (in men) and eriodictyol sulfate, homoeriodictyol glucuronide, and total homoeriodictyol (in women). These findings support the development of sex-specific or personalized nutritional strategies to maximize the health benefits of functional beverages. The results also highlight the importance of including sex as a biological variable in nutritional and clinical trials, as male and female metabolisms process these compounds differently. The clustering analysis revealed distinct metabolic phenotypes at baseline and after intervention, suggesting that individual variability in metabolic pathways may also play a role. Future research should explore the underlying mechanisms of sex-dependent metabolic pathway regulation and the potential for stevia as a bioavailability enhancer in functional food formulations.