In silico modeling of pharmaceutical formulation using artificial neural networks

The objective of this study was to develop neural network model of drug release from HPMC matrix tablets in terms of formulation factors and process variables. The physicochemical properties of the drug and HPMC and manufacturing process were investigated and used as independent factors. The % cumulative release of different drugs from hyroxypropylmethylcellulose (HPMC) matrix tablets was used as the response factors. The correlation between causal factors and response factor was examined using feed-forward back-propagation neural networks. The in silico model was optimized by considering goodness-of-fit and cross-validated predictability. A “leave-one-out” cross-validation revealed that the neural network model could predict release properties of drug from HPMC tablets with a reasonable accuracy (predictive r2 of 0.73–0.89 and predictive root mean square error of 1.68–8.90). The predictive ability of these models was validated by a set of 3 formulations that were not included in the training set. The predicted and observed cumulative releases (%) were well correlated.