Ultrasound-assisted extraction of anti-inflammatory actives from corn silk (Zea mays L.): process optimization, machine learning screening, and interaction mechanisms

Corn silk ( Zea mays L. ), an accessory product of maize, is an unconventional component with potential bioactive compounds. In this study, isoorientin was extracted from maize whiskers by ultrasound-assisted extraction, and the response surface optimization resulted in an extraction amount of 1.26 mg/g. A quantitative structure-effect relationship (QSAR) prediction model was developed, and the R 2 and RMSE of the XGBoost regression algorithm were 0.9582 and 0.01823, respectively. Interpretative analyses of the prediction model were carried out by using the SHAP visualization, which showed that the charge, structure, and lipid solubility of the molecules play a key role. Several anti-inflammatory compounds were screened by combining molecular docking and molecular dynamics. After oral toxicity testing, it was determined that isoorientin performed the best and had a semi-inhibitory concentration of 18.38 μg/mL. This study improves the high-value utilization of maize whiskers and provides a theoretical basis for the rapid screening of food-borne anti-inflammatory compounds.