Modeling and optimization of a pharmaceutical formulation system using radial basis function network.

Area of Science:

• Pharmaceutical Science
• Computational Chemistry
• Chemical Engineering

Background:

• Pharmaceutical formulation involves complex interactions between formulation factors and process variables.
• Optimizing formulations with multiple objectives presents challenges due to conflicting property requirements.
• Existing methods like Response Surface Method (RSM) may not efficiently handle multi-objective optimization problems.

Purpose of the Study:

• To propose a novel Radial Basis Function Network (RBFN) based method for modeling and optimization of pharmaceutical formulations.
• To address the challenge of conflicting objectives in multi-objective pharmaceutical formulation.
• To evaluate the performance of the proposed RBFN method against traditional techniques.

Main Methods:

• Development of a novel Radial Basis Function Network (RBFN) model for pharmaceutical formulation.
• Implementation of a hierarchically self-organizing learning algorithm for automatic RBFN configuration.
• Evaluation using a trapidil formulation system and comparison with Response Surface Method (RSM) based on multiple regression.

Main Results:

• The proposed RBFN method demonstrated superior performance in modeling and optimization compared to regression-based RSM.
• The RBFN method effectively handled multiple objectives in the pharmaceutical formulation system.
• Automatic configuration of RBFN parameters was achieved through the hierarchical learning algorithm.

Conclusions:

• The novel RBFN method offers an efficient approach for multi-objective pharmaceutical formulation modeling and optimization.
• RBFN provides a robust alternative to traditional methods like RSM for complex formulation challenges.
• This technique facilitates the development of optimal pharmaceutical formulations by effectively managing conflicting objectives.