Prediction of a stable microemulsion formulation for the oral delivery of a combination of antitubercular drugs using ANN methodology

Purpose: The aim of this project was to develop a colloidal dosage form for the oral delivery of rifampicin and isoniazid in combination with the aid of artificial neural network (ANN) data modeling. Methods: Data from the 20 pseudoternary phase triangles containing miglyol 812 as the oil component and a mixture of surfactants or a surfactant/cosurfactant blend were used to train, test, and validate the ANN model. The weight ratios of individual components were correlated with the observed phase behavior using radial basis function (RBF) network architecture. The criterion for judging the best model was the percentage success of the model prediction. Results: The best model successfully predicted the microemulsion region as well as the coarse emulsion region but failed to predict the multiphase liquid crystalline phase for cosurfactant-free systems indicating the difference in microemulsion behavior on dilution with water. Conclusions: A novel microemulsion formulation capable of delivering rifampicin and isoniazid in combination was created to allow for their differences in solubility and potential for chemical reaction. The developed model allowed better understanding of the process of microemulsion formation and stability within pseudoternary colloidal systems