Statistical Modeling for the Optimization of Bioluminescence Production by Newly Isolated Photobacterium sp. NAA-MIE

Abstract

The utilization of bioluminescent bacteria has served as a powerful bioassay in environmental monitoring as a result of simplicity of use, quick response and high sensitivity on the basis of the luminescence being produced. In the present study, the optimization process for the luminescence production of a newly isolated Photobacterium sp. NAA-MIE using response surface methodology (RSM) and artificial neural network (ANN) is investigated. RSM and ANN are the most favored techniques and efficient methods for the optimization of medium components, especially for nonlinear systems. This analysis presents the comparative evaluation between RSM and ANN aimed at their particular empirical modeling and predictive ability. RSM predicted the optimized condition with four major variables at glycerol concentration of 0.16%, NaCl of 2.49%, pH at 7.4 and tryptone of 2.97% with 260,748.94 RLU of the maximum luminescence production. The accuracy of the RSM model equations was measured with R2 of 0.9001 and adjusted R2 of 0.8001. ANN predicted optimum condition at a glycerol concentration of 0.10%, NaCl of 2.46%, pH at 7.5 and tryptone of 1.97% with 237,481.32 RLU. The study demonstrated that ANN accomplishes more than RSM with a higher R2 value of 0.968 and low mean percentage error and root mean square error, which were 0.003 and 3.66, respectively. Both models are presented as suitable predictive models on the basis of mathematical modeling to optimize and improve biological systems for future upscaling processes with ANN as more predictive and having more fitting potentiality compared to RSM.