QSAR Modeling of Some Derivatives of Thiazolidinedione With Antimalarial Properties

Abstract

Malaria is a serious human health threat that affects the lives of millions of people annually. To this end, the Quantitative structure–activity relationship (QSAR) of 31 thiazolidinedione derivatives were used to predict anti-malarial activity. Multiple linear regression (MLR) model and artificial neural network (ANN) are used for modeling. The best results were obtained for thiazolidinedione derivatives with 5 descriptors. The obtained results indicated that the MLR implemented for thiazolidinedione derivatives with parameters: R²: 0.90, R²_adj: 0.88, Q²: 0.89, and RMSE: 2.06. Also, the ANN was used in which the correlation coefficients of the three groups of train, validation, test and total were 0.94, 0.98, 0.99, and 0.95, respectively. Based on the results, a comparison of the quality of the models show that the ANN model has a significantly better predictive capability. ANN establishes a satisfactory relationship between the molecular descriptors and the activity of the studied compounds.