Abstract Skin aging involves complex molecular changes that current strategies struggle to reverse. Here, we developed a machine learning approach using Support Vector Regression to predict biological skin age from proteomic profiles, enabling objective assessment of anti-aging formulations. In this exploratory, proof-of-concept, prospective observational study, we set out to characterize treatment-induced proteomic changes in human skin as the pre-specified primary outcome. Women (ages 20–80) received 30-day topical quinoa bioester application on one forearm, with the contralateral forearm receiving vehicle. Mass spectrometry revealed significant upregulation of barrier function proteins (desmoglein-1, filaggrin), antioxidant enzymes (SOD1, glutaredoxin-1), and protease inhibitors. Our Support Vector Regression (SVR) model, trained on pre-treatment proteomes, predicted lower proteomic ages for quinoa bioester-treated skin compared to vehicle-treated skin, with observed median differences of 11 and 16 years for participants under and over 50, respectively (p < 0.01 for participants ≥50 years). While these values do not necessarily correspond to biological years, these findings demonstrate that topical bioactives can induce detectable shifts in skin proteomic profiles. These results establish a quantitative framework for evaluating skin rejuvenation strategies and suggest quinoa bioester as a promising anti-aging cosmeceutical.
Camillo-Andrade et al. (Thu,) studied this question.