Machine learning insights into the oxidative and cytotoxic effects of Ceratophyllum demersum extracts on HepG2 and THLE2 cells

Abstract This study evaluates the cytotoxic, antioxidant, and oxidative effects of methanol and water extracts of Ceratophyllum demersum on hepatocellular carcinoma (HepG2) and normal hepatocyte (THLE2) cell lines. Lactate dehydrogenase (LDH) release assays revealed dose-dependent cytotoxicity in both cell lines. IC₅₀ values for the water extract were calculated as 95.29 µg/mL for HepG2 and 165.44 µg/mL for THLE2. For methanol extract, IC₅₀ values were 196.82 µg/mL (HepG2) and 271.87 µg/mL (THLE2). Total antioxidant status (TAS) levels peaked at low doses (12.5–50 µg/mL). Conversely, total oxidant status (TOS) levels rose sharply, with maximum oxidative stress observed at 200 µg/mL water extract in HepG2 cells. Machine learning analyses using random forest regression, gradient boosting regression, and extreme gradient boosting identified extract concentration and TOS as key predictors of LDH levels, with SHapley Additive exPlanations analysis supporting these findings. Machine learning revealed strong oxidative stress–cytotoxicity links, aiding phytochemical bioactivity analysis.