Integration of multiple linear regression, principal component analysis, and hierarchical cluster analysis for optimizing dried fingerroot (Boesenbergia rotunda) extraction process

Fingerroot (Boesenbergia rotunda) is a medicinal plant. Recently, it was reported to have the highest potent anti-SARS-CoV-2 activity among 122 Thai medicinal plants, owing to its phenolic compounds. In this study, we aimed to optimize the conditions for extracting phenolics and their functional properties from dried fingerroot. To design the extraction conditions, fifteen treatments were obtained from a combination of three independent variables (temperature, time, and methanol content in acetone) using a Box-Behnken design. The extraction conditions were evaluated based on total phenolic content (TPC), ferric ion reducing antioxidant power (FRAP), and acrolein scavenging ability (ACSA). These values were fitted to a quadratic polynomial model utilizing multiple linear regressions (MLR). Principal component analysis (PCA), together with hierarchical cluster analysis (HCA), was then used to select the optimal conditions. The predictive models well described the TPC and ACSA. Employing the optimized conditions, i.e., 45 degrees C, 60 min, and 75% methanol, resulted in the extract having 2.98 mg GAE g(dw)(-1), 2.02 mg TE g(dw)(-1), and 0.156 nmol s(-1)g(dw)(-1) for TPC, FRAP, and ACSA, respectively. These results were 3.5-, 4.1-, and 2.5-fold higher than the lowest values predicted by the developed models for TPC, FRAP, and ACSA, respectively. The extraction conditions for TPC, FRAP, and ACSA from dried fingerroot were successfully optimized. The combined technique (MLR+PCA+HCA) proposed in this study yielded results comparable to those obtained using conventional techniques. Therefore, it can be used as an alternative optimization method.