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

What if the key to unlocking a fruit's hidden antioxidant power lies in the perfect sonic vibration? A new study on Pilangkasa fruit reveals how ultrasound can efficiently extract its valuable compounds—and machine learning can predict the process.

Pilangkasa fruit (Ardisia elliptica Thunb) is a treasure trove of bioactive compounds, yet its potential has been largely overlooked. Now, researchers are turning to green extraction methods—specifically ultrasound-assisted extraction—to efficiently recover these valuable antioxidants. The study tested five sonication amplitudes, from 0% to 100%, over time periods up to 90 minutes, measuring extraction yield, total phenolic content, and total anthocyanin content.

The results were striking: the power of sonication dramatically influenced how well phytochemicals were recovered. Extraction was fastest in the initial 30 minutes, then leveled off or slightly declined. To make sense of the data, the team fitted five empirical models, with the first-order model emerging as the best match between actual and predicted outcomes. Meanwhile, an artificial neural network (ANN) provided fast and accurate predictions of the extraction process.

Based on the extraction rate constants, amplitudes of 50–75% proved optimal for extracting polyphenols with high yields and high levels of total polyphenols and anthocyanins. These conditions offer a promising path for scaling up in the food industry, turning a humble fruit into a powerhouse of natural antioxidants.

Key Points

• Ultrasound amplitude greatly boosts phytochemical extraction efficiency.
• First-order model best predicts extraction kinetics among five tested.
• 50–75% amplitude optimal for high polyphenol and anthocyanin yields.