A statistical and neural network-assisted sustainable integrated process-based on 'zero solid waste' for the extraction of polyphenols, dietary fiber and xylooligosaccharide from de-oiled rice and corn bran.

Response surface methodology (RSM)-Box–Behnken design was used to optimize recovery of polyphenols, dietary fiber and xylooligosaccharide (XOS) from rice and corn bran in an integrated process. Same parameters were used to train artificial neural network (ANN) model with Levenberg–Marquardt algorithm. ANN showed better prediction accuracy as compared to RSM due to higher coefficient of determination (R2). Optimized conditions for extraction of polyphenols from rice bran were 70% ethanol, 1:10 solvent to sample ratio at 75 °C for 12 h. Moreover, extraction of corn bran polyphenols was optimized at 80% ethanol, 1:10 solvent to sample ratio at 73.6 °C for 10.5 h. After the extraction of polyphenol, left-over bran residue was subjected for dietary fiber extraction. Further, conditions were also optimized for the extraction of XOS from rice bran dietary fiber and found to be pH 6.92, sample to buffer ratio 1:11 with 243.86 units of xylanase enzyme at 46.77 °C for 5.23 h. In case of corn bran dietary fiber XOS, optimization conditions were pH 5.09, sample to buffer ratio 1:12 with 229.12 units of xylanase enzyme at 53.22 °C for 8.02 h. The optimized conditions gave a yield of 8.5% and 9% of polyphenols; followed by 45.1% and 49.2% of dietary fiber and later on 32% and 33.9% of XOS from rice and corn bran, respectively. All extracted bioactive compounds were found to be of good quality. Hence, an integrated process developed at gram scale of rice bran and corn bran was also found to be energy-efficient and cost-effective. [ABSTRACT FROM AUTHOR]