Modeling the green extraction of bioactive compounds from Pilangkasa fruit (Ardisia elliptica Thunb) through empirical and machine learning approaches

Abstract

Pilangkasa Fruit (Ardisia elliptica Thunb) contains the rich bioactive compounds, which still limited the study to utilization. This study aimed to investigate the effect of green extraction technique – ultrasoud-assisted extraction on the antioxidants recovery efficient and establish the kinetic extraction under different conditions. Five level of sonication amplitute (0, 25, 50, 75, 100%) were used for extraction at a time range from 0 to 90 minutes. The extraction yield (%), total phenolic compound (mg gallic acid equivalent per g, mgGAE/100 g), total anthocyanin content (mg/100 g) were determined. The experimental data was fitted with five empirical models to find the best fit model. The power of sonication greatly influenced the efficiency for recovering the phytochemical compounds in Pilangkasa fruit (Ardisia elliptica Thunb). The fast rate of extraction was found at the initial stage (after 30 min of extraction), then reached the equilibrium stage or slightly declined. Among the five empirical models, first-order model showed the best fit between the actual and predicted data. While, ANN model presented the fast and accurate prediction the extraction process. Based on the extraction rate constant from the first-order model, the level of around 50-75% of amplitute was considered as the appropriate condition for extraction polyphenol from Pilangkasa fruit with high yield of extraction, total polyphenol content, total anthocyanin content. These conditions could be further optimized and upscaled for used in food industry.