Artificial Neural Network (ANN) Approach to Predict an Optimized pH-Dependent Mesalamine Matrix Tablet

Severe bleeding and perforation of the colon and rectum are complications of ulcerative colitis which can be treated by a targeted drug delivery system.

Development of colon-targeted delivery usually involves a complex formulation process and coating steps of pH-sensitive methacrylic acid based Eudragit^®. The current work was purposefully designed to develop dicalcium phosphate (DCP) facilitated with Eudragit-S100-based pH-dependent, uncoated mesalamine matrix tablets.

Mesalamine formulations were compressed using wet granulation technique with varying compositions of dicalcium phosphate (DCP) and Eudragit-S100. The developed formulations were characterized for physicochemical and drug release profiles. Infrared studies were carried out to ensure that there was no interaction between active ingredients and excipients. Artificial neural network (ANN) was used for the optimization of final DCP-Eudragit-S100 complex and the experimental data were employed to train a multi-layer perception (MLP) using quick propagation (QP) training algorithm until a satisfactory root mean square error (RMSE) was reached. The ANN-aided optimized formulation was compared with commercially available Masacol^®.

Compressed tablets met the desirability criteria in terms of thickness, hardness, weight variation, friability, and content uniformity, ie, 5.34 mm, 7.7 kg/cm^2, 585±5 mg (%), 0.44%, and 103%, respectively. In-vitro dissolution study of commercially available mesalamine and optimized formulation was carried out and the former showed 100% release at 6 h while the latter released only 12.09% after 2 h and 72.96% after 12 h which was fitted to Weibull release model with b value of 1.3, indicating a complex release mechanism.

DCP-Eudragit-S100 blend was found explicative for mesalamine release without coating in gastric and colonic regions. This combination may provide a better control of ulcerative colitis.