Predictive reducing sugar release from lignocellulosic biomass using sequential acid pretreatment and enzymatic hydrolysis by harnessing a machine learning approach

The development of sustainable production of bio-based chemicals is targeted towards the use of lignocellulose. Overcoming its recalcitrance is a crucial step for biomass valorization. Here, we demonstrate a predictive system for reducing sugar yield from acid pretreatment and enzymatic hydrolysis processes of two types of biomass, rice straw and sugarcane leaves, which have different lignocellulosic compositions. A machine learning model based on the Decision Tree algorithm was employed to predict the amount of reducing sugars generated during enzymatic hydrolysis. The model demonstrated satisfactory accuracy, with an R² of 0.8910 for the training set and 0.8121 for the testing set, along with low error values (RMSE 0.1042 and MAE 0.0705). Scanning electron microscope (SEM) revealed that the biomass structure undergoes significant changes after enzymatic hydrolysis, as proven by the formation of surface pores. This morphological alteration reflects the enzymatic degradation of cellulose, resulting from the disruption of fiber bonds. The application of machine learning in this research shows great potential for enhancing biomass conversion efficiency, contributing to biomass valorization efforts.