INTRODUCTION

Orally disintegrating tablets (ODTs) are popular due to their rapid disintegration, improved patient compliance, and enhanced therapeutic efficacy. Designing ODTs requires careful optimization of formulation variables to achieve desired quality attributes. This study integrates Central Composite Design (CCD) and Artificial Neural Networks (ANN) to optimize Doxylamine Succinate ODTs using binder and superdisintegrant variables.

METHODS

CCD was used to design and develop directly compressed Doxylamine Succinate ODTs. Binder concentration (Povidone) and superdisintegrant (Crospovidone) were the independent variables, with tablet friability, wetting time, and disintegration time as the dependent variables (responses). The compressed tablets were evaluated for all the responses. The dissolution studies were also performed in Hydrochloric Acid (0.01N). Response data were used to train the ANN-based model to optimize the formulation. ANN-based optimized formulation was assessed for drug release, and the release profiles of CCD and ANN-based optimized formulations were compared, using the f₂ test (similarity factor test).

RESULTS

All the CCD proposed formulations showed appropriate wetting time (10-12 seconds) with disintegration time ranging from 27-29 seconds and friability lower than 1%. All the formulations showed drug release above 80%. The highest drug release was observed with F7, which was considered the optimized formulation based on CCD results.

The selection of the optimal number of nodes influences the performance of the ANN model. The model for the current study was trained using the TanH values ranging from 3 to 10. The SSE and r² values were recorded at each node value. Node value 5 was the best activation node. The release profile of the ANN-optimized formulation was also assessed. The f₂ test demonstrated similarity between the two formulations.

CONCLUSION

The integration of ANN with CCD provides a robust, reliable, and multi-objective optimization platform for ODT development.