Bioengineering for multiple PAHs degradation using process centric and data centric approaches

Abstract

The study aims sequential bioengineering for multiple PAHs degradation using a process centric approach – response surface methodology (RSM) and a data centric approach – artificial neural networks (ANN). The study involves stepwise media optimization protocol for multiple PAHs degradation using newly isolated Stenotrophomonas maltophilia. RSM, a non linear model has predicted 94.70% degradation on 5th day with R2 value of 0.97. The analysis of desirability revealed the optimum value of the process conditions: NaCl – 27.55 g/L; NH4Cl- 0.28 g/L; Na2HPO4- 0.08 g/L, TAPSO- 1.74 g/L. Feed forward ANN, a linear model has predicted 95.94% degradation with R2 value of 0.99. The change in the magnitude of error functions viz. root mean square error (RMSE), mean absolute deviation (MAD) and mean absolute percentage error (MAPE) were low during ANN prediction as compared to RSM. Moreover, sensitivity analysis of both models also proves efficiency of the prediction capability and generalization of the data. Thus, for the very first time chemometrics study of medium components for multiple PAHs degradation offers constructive and powerful alternative to scientific community to design microcosm and mesocosm experiments.