Machine learning modeling and response surface methodology driven antioxidant and anticancer activities of chitosan
Seyma Bulut1, Muhammad Aasim2, Bugrahan Emsen2
1Department of Biotechnology, Faculty of Science, Necmettin Erbakan University, 42090 Konya, Turkey.
Abstract:
This study investigates the potential of chitosan nanoparticles (CNPs) in enhancing the bioavailability and efficacy of Bacopa monnieri extracts, known for their neuroprotective, antioxidant, and anticancer properties. Different concentrations of CNPs were added to the culture medium for in vitro shoot regeneration. Antioxidant activity (DPPH free radical scavenging and H2O2 removal assays) and cytotoxicity assay (LDH release and XTT viability) were performed. The results demonstrated the highest DPPH radical scavenging activity of 95.60 % at 125 microg/mL CNPs from methanol extract. Whereas, H2O2 scavenging activity increased with higher extract concentrations, and the maximum was recorded from methanol extract when used at 1000 microg/mL. Cytotoxicity assays revealed a dose-dependent increase in LDH activity and XTT reduction, and water-based extracts demonstrated the strongest cytotoxic effects. IC50 analysis indicated that CNP-enriched methanol and water extracts were significantly more cytotoxic to HeLa cells as compared to ethanol extracts. Response surface regression analysis and ML models confirmed the reliability of the experimental data, with the multilayer perceptron (MLP) model exhibiting the best predictive accuracy, followed by the random forest (RF) model. It can be concluded that CNP enrichment significantly improved the antioxidant and anticancer properties of B. monnieri extracts, highlighting the potential of CNP-based formulations for future studies.


