Introduction: Plant extracts contain a wide variety of phytochemicals that provide valuable biological activities, notably antioxidant properties. Enhancing antioxidant capacity offers significant advantages for applications in healthcare, food, and cosmetics. The pursuit of process optimization has evolved with the advent of artificial intelligence (AI), offering unprecedented opportunities to obtain highly antioxidant extracts. In this study, AI-driven methodologies were applied to optimize antioxidant extraction from Cistus salviifolius while promoting more sustainable processes.
Methods: Optimization of aqueous C. salviifolius extracts was achieved using AIReviewer, a validated free AI-based tool for scientific literature analysis (https://doi.org/10.3390/antibiotics12020327), and a Python-coded Jupyter Notebook for data analysis and response surface methodology optimization via a Box—Behnken design. Antioxidant capacity was measured using TEAC and FRAP assays and total phenolic content by he Folin—Ciocalteu method. This optimization method is also compatible with the results of any antioxidant capacity quantification method including ORAC or DPPH, among others.
Results: AIReviewer literature analysis helped us to identify key extraction variables and their suitable ranges: time (0–240 min), temperature (25–50 °C), and ultrasonic energy (50–150 J/mL). Then, fifteen extractions were performed combining different variables using a Box—Behnken design. The optimal conditions (240 min, 25 °C, and 89.65 J/mL) predicted the maximum antioxidant capacity, yielding experimental results of 510.21 ± 24.69 mmol Trolox equivalents/100 g extract (TEAC) and 804.46 ± 22.11 mmol FeSO₄ equivalents/100 g extract (FRAP). The total phenolic content was 31.42 ± 1.45%, with a 30.11 ± 1.25% yield. These values surpass those reported in the literature, even for extracts obtained with non-aqueous solvents like ethanol.
Conclusion: AI-based optimization techniques effectively enhanced the antioxidant properties of C. salviifolius extracts, producing a phenolic-rich extract with superior antioxidant capacity. These methods also offer a more sustainable approach to extraction by significantly reducing the number of empirical trials needed to determine optimal conditions. The results demonstrate the value of AI in optimizing extraction processes, improving both the quality and sustainability of bioactive product development.
