Empirical Model Variability: Developing a new global optimisation approach to populate compression and compaction...

Area of Science:

• Pharmaceutical Sciences
• Materials Science

Background:

• Empirical models are crucial for predicting pharmaceutical powder compaction behavior.
• Existing models often show reduced accuracy with increasing active pharmaceutical ingredient (API) concentrations.
• Optimization of model parameters is essential for reliable predictions in tablet formulation.

Purpose of the Study:

• To systematically evaluate the predictive accuracy of common empirical models for pharmaceutical powder compaction.
• To develop and validate a novel global optimization approach for empirical compaction models.
• To reduce experimental requirements and API consumption during new blend formulation.

Main Methods:

• Fitting nine placebo and twelve API-loaded blends to Gurnham, Heckel, Kawakita, Ryshkewitch-Duckworth, and Leuenberger models.
• Performing parameter variability analysis to identify acceptable model fits.
• Developing a global optimization approach using arithmetic, geometric, and harmonic mixture rules for tuning parameters.
• Conducting cross-validation studies to assess predictive capability for new blends.

Main Results:

• All models performed well (R² > 90%, RRMSE < 0.1) at low API loadings (<20w/w%).
• Model performance, particularly the Heckel model, decreased significantly with increased API loading.
• The novel global optimization approach outperformed the traditional line of best fit.
• Cross-validation confirmed the method's ability to predict tuning parameters for acceptable Goodness of Fit in new blends.

Conclusions:

• The predictive accuracy of empirical compaction models is sensitive to API loading.
• The developed global optimization method offers improved prediction and reduced experimental burden for pharmaceutical blend development.
• This approach enables efficient exploration of new blends with minimized API usage and experimental effort.