Machine Learning Regression for Assessing Sensing Performance and Anticancer Potential of Oolong Tea-Derived Silver Nanoparticles

Abstract

In this study, machine learning (ML) algorithms were employed to predict analyte concentrations using sensing results and evaluate the anticancer effects of nanostructures. Multifunctional oolong tea extract-mediated silver nanoparticles (OTE-Ag NPs) were synthesized via a photo/ultrasound method and utilized in various applications, including a smartphone-based H2O2 sensor and electrochemical sensors for urea and fructose. Key features were extracted from electrochemical results, and feature importance analysis was used to select the most predictive features. The artificial neural network (ANN) model provided accurate predictions, particularly strong for urea (R2 = 0.8575, RMSE = 0.4266, MAE = 0.3380). The study revealed the selective toxicity of OTE-Ag NPs to MCF-7 breast cancer cells through analyses of cytotoxicity, apoptosis, cell cycle phases, and CD44 surface marker expression using Annexin V/PI dye and flow cytometry. Experimental results demonstrated that OTE-Ag NPs suppressed MCF-7 cell proliferation while exhibiting lower cytotoxicity in normal HUVEC cells (46% cell death). OTE-Ag NPs arrested MCF-7 cells in the G2/M phase, induced apoptosis, and reduced CD44 expression, suggesting metastasis suppression. The CD44+/CD24- ratio decreased from 84.79% in control MCF-7 cells to 47.7% in OTE-Ag NP-treated cells. Overall, OTE-Ag NPs significantly inhibited MCF-7 cell proliferation through the apoptotic pathway by regulating the cell cycle in the G2/M phase.