Within the framework of green chemistry and wild fruit valorization, this study optimizes the extraction of bioactive compounds from Prunus spinosa L. fruits using glycerin-based ultrasound-assisted extraction (UAE). Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) were comparatively employed to model the process. Significant improvements in extraction efficiency were achieved, with total phenolic content increasing from 9.28 to 23.22 mg GAE/g DW, total flavonoid content from 6.53 to 21.65 mg CE/g DW, and antioxidant activity (DPPH assay) from 57.04% to 86.34%. While both models performed well, ANN demonstrated slightly higher predictive accuracy, highlighting its potential for capturing complex, non-linear relationships in the extraction process. We identified the optimal extraction conditions as 9 min extraction time, 100% ultrasonic amplitude, and 40% water in glycerin, and these conditions were experimentally validated. UHPLC-DAD-HRMS/MS profiling revealed a rich phytochemical fingerprint dominated by phenolic acids, caffeoylquinic acid derivatives, and flavonol glycosides, and revealed largely overlapping qualitative phytochemical profiles between hydroglyceric and ethanolic extracts. Comparative extraction using 70% ethanol under identical conditions resulted in lower TPC, TFC, and antioxidant activity, indicating the improved efficiency of glycerin under the investigated conditions. Overall, the optimized glycerin-based UAE provides a sustainable, food-safe approach for extracting bioactive compounds from underutilized P. spinosa fruits. These results support its application in functional foods and in nutraceutical and cosmetic formulations.
Berkati et al. (2026) studied this question.