Abstarct: The aim of this project was to employ an artificial neural network (ANN) modeling approach combined with the selection of the most effective descxriptors using the Random Forest (RF) method. The RF variable selection method identified 10 descxriptors that had the highest relevance to the pharmaceutical activity of the compounds. Subsequently, these selected descxriptors were used as inputs to an artificial neural network to develop a QSAR model. For the ANN model, the dataset was divided into three sets: training, validation, and testing. With the goal of constructing an optimal ANN model with suitable performance, all influential network parameters were optimized and trained. After developing the optimal model, the pharmaceutical activity of the compounds in the test set was predicted. The coefficient of determination (R²) for the test data in the RF-ANN model was 0.81. The developed model was evaluated using various statistical methods, including stepwise removal of individual descxriptors, applicability domain analysis, Y-randomization, and calculation of a series of statistical parameters. The evaluation results demonstrate the generalizability and predictive power of the models developed in this study