The application of "artificial neural network" in dosage form development of bi-layer floating tablets of baclofen

The aim of the present investigation was to prepare & evaluate bilayer floating tablet of Baclofen. In which one layer was made up of immediate release layer to provide loading dose, while another layer was made up of floating layer of BCF to provide maintenance dose. The focus of present work was to prepare and evaluate bilayer floating tablet of Baclofen to increase residence time in stomach and there by gives prolog action. Tablets were prepared by wet granulation technique. Drug-excipients compatibility study was done by using Fourier transform infrared spectroscopy (FTIR). Optimization was carried out using artificial neural network (ANN) and multiple regression analysis using 3 factorial designs. Cumulative percentage release (CPR) at 24 hr, time required for 50% of drug release (T50%), floating lag time study of the tablet formulations were selected as dependent variables. The Content of HPMC K 4M (X1) and Content of PEO WSR N 10 (X2) were selected as independent variables. Tablets were evaluated for swelling index, in vitro buoyancy and in vitro drug release. The similarity factor (f2) was used as a base to compare dissolution profiles. Optimized batch was subjected for kinetic modeling. Different process parameters of optimized batch were also studied. From FTIR spectra it was observed that there were no any interaction between drug and excipients used. The results demonstrate that 3.5% of crosspovidone released 99% of drug in 20 minutes. It was found that HPMC K 4M with concentration 30% and PEO WSR N 10 with concentration 20% showed good sustained as well floating ability and its releases 99.39% of drug within 24 hrs. Drug release was best explained by Higuchi plot. It was seen that the process parameters have great influence on performance of bilayer floating tablet. To check the accuracy of these predictions, experimentally three formulations were prepared by random selection of causal factors as per counter plot and also validated ANN. The experimental data were compared with predicted data by paired t test, no statistically significant difference was observed. ANN showed less error compared with multiple regression analysis. These findings demonstrate that ANN provides more accurate prediction and was quite useful in the optimization of pharmaceutical formulations than the multiple regression analysis method.