Data-driven interpretable analysis for polysaccharide yield prediction.

Area of Science:

• Biotechnology
• Agricultural Science
• Data Science

Background:

• Cornstalks are a promising source for polysaccharide production using xylanase.
• Accurate prediction of polysaccharide yield is crucial for optimizing enzymatic processes and reducing costs.
• Enzymatic factor interactions complicate precise yield prediction and optimization.

Purpose of the Study:

• To develop a data-driven approach using artificial intelligence for enhanced polysaccharide production from cornstalks.
• To identify accurate machine learning models for predicting polysaccharide yield.
• To uncover optimal enzymatic parameter combinations for maximizing yield.

Main Methods:

• Implementation of a machine learning framework incorporating Random Forest (RF), eXtreme Gradient Boost (XGB), and deep neural network (DNN) models.
• Utilizing feature importance analysis to identify key enzymatic parameters influencing polysaccharide yield.
• Applying interpretability analysis to understand complex parameter interactions.

Main Results:

• XGB model achieved the highest prediction accuracy (95.6%), followed by RF (93.0%) and DNN (91.1%).
• Enzyme solution volume (43.7%) was identified as the most significant factor, followed by time, substrate concentration, temperature, and pH.
• Identified complex parameter interactions and potential optimization strategies.

Conclusions:

• A data-driven, AI-powered approach effectively predicts polysaccharide yield from cornstalks.
• Machine learning models, particularly XGB, offer robust solutions for process optimization.
• This methodology facilitates efficient recovery of polysaccharides from agricultural residues.