Crystallinity and polymorphism changes during pharmaceutical manufacturing can impact drug quality, manufacturability, and patient safety. While traditional characterization methods exist, real-time monitoring remains challenging. In a study published on October 20, 2025, in Frontiers of Chemical Science and Engineering , researchers from the U.S. Food and Drug Administration illustrate how soft sensors can be used to predict critical quality attributes (CQAs) and track solid-state changes in key unit operations.
The work focuses on drugs with moderate to high solid-state risk levels and applies models including population balance models, semi-empirical models, and statistical correlations to wet granulator, fluidized bed dryer, mill, and tablet press operations. Sensitivity analysis is used to quantitatively assess the impact of process inputs on outputs, supporting risk-based control strategy development.
In the wet granulation case, a population balance model predicted particle size distribution, while a smoothing splines model captured polymorphic transformation as a function of liquid-to-solid ratio. For fluidized bed drying, a Midilli-Kucuk model predicted product temperature and moisture content. In milling, a population balance model tracked particle size reduction, and an exponential model described crystallinity loss over time. For the tablet press, statistical correlations were developed between compression pressure and both tablet tensile strength and polymorphic form change.
Sensitivity analysis using the partial rank correlation coefficient method identified critical process parameters that most strongly influence quality attributes. For example, in wet granulation, liquid-to-solid ratio was highly correlated with polymorphic transformation, while impeller speed and wet massing time affected particle size.
The study also discusses regulatory considerations, noting that soft sensors align with FDA initiatives in advanced manufacturing, quality-by-design, and process analytical technology. Although no specific guidance on soft sensors exists, their use is consistent with broader frameworks for model-based control, data integrity, and risk management.
The authors note that future work could integrate multiscale data and explore applications in continuous manufacturing. Broader adoption will require collaboration across industry, academia, and regulators, particularly in areas of model validation, lifecycle management, and workforce training.
Frontiers of Chemical Science and Engineering
Experimental study
Not applicable
Soft sensors to predict critical quality attributes and monitor crystallinity and polymorphism change in solid oral dosage manufacturing: case studies
5-Dec-2025