Work

Advanced Hybrid RSM-ANN-GA Modeling for Efficient Ultrasound-Assisted Hot-Water Extraction of Polyphenol-Rich Bioactives from Pilangkasa Fruit

Year: 2026
Type: article
Abstract: Pilangkasa (Ardisia elliptica Thunb.) is a source of natural antioxidants with potential applications. Ultrasound-assisted extraction (UAE) has been proposed as a sustainable alternative to conventional extraction methods, but extraction efficiency highly depends on processing conditions. Therefore, the present study was carried out to optimize the UAE of phytochemicals from polyphenol-rich fruit using Response Surface Methodology (RSM) and Artificial Neural Network-Genetic Algorithm (ANN-GA) modeling. A Box-Behnken design was used to study the effects of extraction temperature (50 to 70 °C), time (10 to 30 minutes), and ultrasonic power (50 to 70%) on the total phenolic content (TPC), total flavonoid content (TFC), and antioxidant activities (DPPH and FRAP). The RSM models were highly predictive with coefficients of determination ranging from 85.52 to 98.71%. Among all the responses, extraction time was found to be the most influential factor. Results indicated that cavitation-enhanced mass transfer was the main factor influencing TPC and TFC, whereas antioxidant activity was greatly influenced by structural stability and the sonochemical degradation of polyphenols. The hybrid RSM-ANN-GA predicted the optimal conditions to be 60.2 °C, 22.45 minutes, and 60.34% ultrasonic power, resulting in TPC of 92.24 mg GAE g-1, TFC of 175.35 mg QE g-1, DPPH of 81.27%, and FRAP of 1.61 mg TE g-1. The ANN-GA model showed better prediction than RSM, with high correlation and slightly improved predicted optimal conditions. These results illustrate the potential of hybrid modeling approaches for optimizing extraction processes and provide insights into the physicochemical mechanisms underlying polyphenol recovery.
Language: English
Field: Medicine
Open Access status: gold
Estimated APC paid: $65
OpenAlex ID: W7203752041
FWCI:
Cites: 40
Cited by:
Related to: 10