(555b) Enhancing the Understanding of Active Pharmaceutical Ingredient Crystallization through Data-Driven Modelling

Extensive research efforts are dedicated to studying and understanding the dynamics of crystallization, aiming to optimize process parameters and enhance product quality and yield. Traditional modeling approaches such as Population Balance Models (PBM), have limitations in accounting for all crystallization mechanisms and require numerical methods for most situations. Data-driven modeling approaches, including Artificial Neural Networks (ANNs) regressions, have been explored to overcome these limitations. However, challenges such as data availability and limitations in specific modeling approaches need to be addressed.

This study employs a data-rich experimentation approach to capture the intricate dynamics of antisolvent crystallization of an Active Pharmaceutical Ingredient (API). Through the use of Process Analytical Technologies (PAT) and reactor automation, a solubility model is derived from experimental data and a chemometric model based on Partial Least Squares (PLS) regression is developed to determine supernatant API concentration in an automated and data-efficient manner. High-quality time-resolved data of supernatant concentration and supersaturation enable the application of data-driven modeling using Dynamic Response Surface Methodology (DRSM). This approach provides a comprehensive understanding of supersaturation dynamics within the explored design space of API crystallization. The DRSM models are developed using antisolvent fraction-resolved data obtained from crystallization runs conducted based on a 2² full-factorial Design of Experiments (DoE). The models provide insights into the sensitivity of the crystallization process to operating parameters and enable rapid process development. The DoE experiments revealed the influence of antisolvent addition rate and temperature on concentration and supersaturation profiles. The DRSM models successfully described the concentration and supersaturation dynamics based on antisolvent fraction, addition rate, and temperature. The study demonstrates the effectiveness of applying DRSM in the context of crystallization and provides valuable insights for process optimization and control in pharmaceutical manufacturing.