Machine learning models predict pH-responsive drug release from liver-targeted agarose-CMC-CeO2 nanocomposites: Liver-targeted antioxidant strike

This research presents an innovative carboxymethyl cellulose/agarose/cerium oxide (CMC/Aga/CeO₂) nanocarrier for the pH-sensitive delivery of quercetin (QC) to hepatocellular carcinoma cells. The nanocarrier exhibited exceptional drug-loading proficiency, achieving 88.25 % entrapment efficiency and 51.00 % loading capacity, which was significantly enhanced by the use of CeO₂. The ideal physicochemical characteristics comprised a consistent hydrodynamic diameter of 193.48 nm and elevated colloidal stability (zeta potential: -61.6 mV). The system exhibited tumor-selective release kinetics, discharging 98 % of QC at an acidic tumor pH (5.4) compared to 64 % at physiological pH (7.4) over a 96-h duration. In vitro cytotoxicity experiments validated precise targeting, leading to a 40 % decrease in HepG2 cell viability and a 94 % preservation of L929 normal cell viability. Alongside these findings, machine learning modeling techniques (Gradient Boosting, Neural Networks, Random Forest, and SVM) were devised to replicate release kinetics, accurately characterizing non-linear pH–time profiles with remarkable precision (R² > 0.97). These prediction methodologies provide computational assistance for enhancing nanocarrier design beyond existing kinetic models. This multi-functional platform effectively mitigates some of the major limitations of conventional QC treatment, including low solubility, systemic toxicity, and non-specific biodistribution, and thereby offers a biocompatible and computationally optimized platform for precision oncology.