Optimizing agarase production from <i>Microbulbifer</i> sp. using response surface methodology and machine learning models

Agarase enzymes are critical in industries like food, cosmetics, and medicine where they play a critical role in DNA recovery, food gelling, cosmetic formulations, and waste treatment. However, current agarase sources often face limitations related to low yields, inconsistent activity, and high production costs. Therefore, there is a need to identify and optimize more efficient microbial sources for industrial-scale agarose production. This study is an exhaustive investigation into the optimized production of extracellular agarase from a microbial source. Through qualitative-quantitative analysis, the study optimizes the growth conditions of <i>Microbulbifer</i> sp. for enhanced agarase production. Response surface methodology is used to investigate the interactive effects of key parameters to get the optimized conditions as 0.3% agar, pH 7, 25°C temperature, and 36-hour incubation time, confirmed by a verification experiment yielding 317.97 μmol min^-1 agarase activity (F-value of 44.75 and an R-squared of 0.9827). The study also explores various machine learning algorithms where radial basis function neural network performed best with R-squared values of 0.989 and low mean squared error of 0.44, indicating the reliability and robustness of predicting agarase activity with high accuracy and generalization. The optimized production conditions and machine learning predictions offer significant improvements in the scalability and efficiency of agarase production with incubation time and temperature having the most dominating effect on agarase production. These findings would help in scaling up production and real-time adjustments during bioreactor operations in an industrial setup.