Data-Driven Optimization of Cellulase-Assisted Extraction of Flavonoids and Chlorophyll from Silkworm Excrement

Guangxi, one of China’s dominant sericulture regions, generates substantial silkworm excrement (SE) annually, yet most remains underutilized. This study optimized cellulase-assisted ethanol extraction of flavonoids and chlorophyll from silkworm frass. Systematic experimentation (n = 31) revealed that extraction temperature exerted dominant influence on both contents (r = 0.54 and 0.37 for chlorophyll and flavonoids, respectively), while the two contents exhibited near-zero correlation (r = 0.06). An XGBoost model achieved R2 = 0.9146 for flavonoid prediction; SHAP analysis identified a critical temperature threshold (~40 °C). Monte Carlo simulation (n = 10,000) constructed a Pareto frontier for multi-objective optimization. The optimized condition (9% enzyme, 50 °C) achieved chlorophyll and flavonoid contents of 1.13 and 6.42 mg/g, respectively. These findings demonstrate that sericulture waste can serve as a biorefinery feedstock and that interpretable machine learning can navigate multi-objective extraction challenges under data constraints.