Original version
Journal of Biophotonics. 2022, 15 (9):e202200097, DOI: https://doi.org/10.1002/jbio.202200097
Abstract
In the process of converting food-processing by-products to value-addedingredients, fine grained control of the rawmaterials, enzymes and process conditionsensures the best possible yield and eco-nomic return. However, when raw mate-rial batches lack good characterization andcontain high batch variation, online or at-line monitoring of the enzymatic reac-tions would be beneficial. We investigate the potential of deep neural networks inpredicting the future state of enzymatic hydrolysis as described by Fourier-trans-form infrared spectra of the hydrolysates. Combined with predictions of averagemolecular weight, this provides a flexible and transparent tool for process moni-toring and control, enabling proactive adaption of process parameters.