Beyond Traditional Methods: Deep-Learning Machines Empower Fingerroot (Boesenbergia rotunda)-Extract Production with Superior Antioxidant Activity

This study investigated the impact of drying parameters on the quality of fingerroot (<i>Boesenbergia rotunda</i>) extract, focusing on phenolic compounds, flavonoids, and antioxidant activity. A Box-Behngen design was employed to evaluate the effects of maltodextrin concentration, inlet temperature, and outlet temperature on the extract's properties. The highest total phenolic content (18.96 µg of GAE/mg extract) and total flavonoid content (33.52 µg of GE/mg extract) were achieved using 20% maltodextrin, a 160 °C inlet temperature, and an 80 °C outlet temperature. Antioxidant activity, measured by DPPH and FRAP assays, was also influenced by drying parameters. Stepwise regression analysis revealed that maltodextrin concentration significantly affected all responses, while the inlet temperature had no significant effect. The outlet temperature significantly influenced FRAP activity. The developed mathematical models accurately predicted experimental values, validating the effectiveness of the RSM and Deep-Learning Machine. Optimal drying conditions for maximizing phenolic compounds were determined to be 20% maltodextrin, a 150 °C inlet temperature, and a 70 °C outlet temperature, resulting in TPC 15.33 µg of GAE/mg extract, TF 28.75 µg of GE/mg extract, IC₅₀ value of 3.99 µg/mL, FRAP value at 4.44 µmoL Fe²⁺/mg extract of phenolic content, and 18.96 µg of the GAE/mg extract. Similar conditions were found to be optimal for maximizing flavonoid content. These findings provide valuable insights for optimizing the drying process of fingerroot extract to preserve its bioactive compounds and enhance its potential applications.