Evaluation of photoantioxidant activities of SnO<inf>2</inf>, doped SnO<inf>2</inf>, and dual-doped SnO<inf>2</inf> using artificial neural networks and neuro-fuzzy system


The purpose of this study is to develop a machine learning model for estimation of photoantioxidant activities of (tin(IV) oxide) SnO2, Co-doped SnO2, Ni-doped SnO2, and Co, Ni-dual-doped SnO2 nanoparticles (NPs) using the experimental data collected in the dark and under visible light conditions. The estimation of photoantioxidant activities enables to assess the ability of SnO2, Co-doped SnO2, Ni-doped SnO2, and Co,Ni-dual-doped SnO2 NPs to scavenge free radicals that might be dangerous to human beings and the environment. Artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) techniques were applied to the experimental data, the estimation models were generated, and their performance results were compared. The robustness of the models was tested by performing multiple simulations, and the performance of these models was assessed using various metrics such as coefficient of determination (R2), root mean squared error (RMSE), mean absolute error (MAE), and mean absolute relative error (MARE). The results showed that both ANN and ANFIS models have strong potential for estimating the photoantioxidant activities of SnO2, doped-SnO2, and dual-doped SnO2 NPs. However, ANN with two hidden layers (R2 = 0.9972, RMSE = 0.0071, MAE = 0.0050, and MARE = 0.0098) is better than ANFIS (R2 = 0.9680, RMSE = 0.0310, MAE = 0.0213, and MARE = 0.0401). Further, a sensitivity analysis was also performed to study the potential effects of variables on the estimation of photoantioxidant activities of pure and doped SnO2. The most sensitive input was the ‘visible light condition’ followed by the dopant variable (Co-doped, Ni-doped, and Co,Ni-dual-doped) and pore size for the estimation of the photoantioxidant activities. Overall, the Ni-doped SnO2 under visible light irradiation shows the best prediction performance for photoantioxidant activities. Thus, this study presents an alternative approach to employ machine learning techniques and develop models based on the experimental data for estimating the antioxidant activities of materials for unknown data, which may be useful for a better understanding of the behavior of nanoparticles in different conditions for future studies.

Materials Today Communications