A non-linear modelling approach to predict the dissolution profile of extended-release tablets

Area of Science:

• Pharmaceutical Sciences
• Drug Delivery Systems
• Computational Chemistry

Background:

• Extended-release (ER) tablet formulations require precise control over drug release kinetics.
• Predicting drug release profiles is crucial for ensuring therapeutic efficacy and patient compliance.
• Traditional linear models often struggle to capture complex interactions influencing drug release.

Purpose of the Study:

• To develop and evaluate a novel non-linear modeling approach for predicting dissolution profiles of ER tablets.
• To compare the performance of artificial neural networks (ANN) against linear regression methods (PLS, MLR).
• To identify key formulation factors affecting drug release from ER tablets.

Main Methods:

• Utilized a full-factorial design to investigate the impact of HPMC/CMC grades, API lubrication, and compression force.
• Applied curve fitting to obtain a first-order dissolution equation and the tablet-specific constant 'k'.
• Employed artificial neural networks (ANN), Partial Least Squares (PLS), and Multiple Linear Regression (MLR) for modeling and prediction.

Main Results:

• Artificial neural networks (ANN) demonstrated superior performance in predicting the tablet-specific constant 'k' compared to PLS and MLR.
• The non-linear ANN model effectively captured complex interactions influencing drug release.
• Dissolution profiles simulated using the ANN-predicted 'k' values showed higher accuracy.

Conclusions:

• Non-linear modeling, particularly ANN, offers a more robust approach for predicting drug release from ER tablets.
• ANN can effectively model the intricate relationships between formulation variables and drug release kinetics.
• This approach enhances the ability to design and optimize ER formulations for predictable drug delivery.