Ultrasonic‑Assisted Extraction of Phalerin from Phaleria macrocarpa:Response Surface Methodology and Artificial Neural Network Modelling

摘要

Phaleria macrocarpa is historically treasured remedy for treating various allergies, infections and health complications. Commercial availability of this plant extract, however, remains limited as conventional phytochemical extraction techniques require prolong extraction time, high consumption of solvents, in conjunction to being energy intensive. Herein, this study aimed to statistically optimize the ultrasonic-assisted extraction (UAE) of phalerin from P. macrocarpa using the Box–Behnken design (BBD) and the predictive capability of this approach was compared to a model derived from artificial neural network (ANN). In the optimization experiment, for only three relevant UAE parameters viz. solvent ratio, extraction temperature and solid-to-solvent ratio were examined, for the response of the highest extraction of phalerin. Under an optimized condition (R~2 = 0.98) [71% methanol, 1:45 solid-to-solvent ratio (g/mL) and extraction temperature of 47°C], a satisfactory amount of 4.26 ± 0.51 mg/g of phalerin was attained. Comparison between the RSM and ANN revealed the latter being a better predictive model and yielded an appreciably higher predictive capability (R~2 = 0.99) in terms of average absolute deviation, AAD (0.24%) versus RSM (AAD = 1.03%).