Current state of machine learning implementation in pharmaceutical process modeling for oral solid dosage forms

Driven by the Food and Drug Administration's Quality-by-Design initiative and the advancements of Industry 4.0, the pharmaceutical industry is transitioning from traditional batch manufacturing to advanced manufacturing. This transition requires reshaping manufacturing by enabling predictive adaptive modeling, real-time process optimization, and real-time quality control. Artificial intelligence (AI) and machine learning (ML) are central to this data-driven paradigm and enable technologies to meet these requirements. In this review, we provide a perspective on the applications of ML in the manufacturing of oral solid dosage forms focusing on unit operations and process analytical technology (PAT). Applications span a variety of areas, including the prediction of critical quality attributes such as granules size distribution and moisture content in wet granulation, the optimization of key process parameters such as temperature and screw speed to achieve target product qualities in extrusion processes, and the development of adaptive, real-time PAT frameworks. While ML demonstrates promise in enabling requirements of advanced manufacturing, challenges remain in data availability, model interpretability, extrapolation, uncertainty quantification, and integration with existing manufacturing workflows. Future directions include integrating ML with digital twins, addressing model scalability to work across different scales, and incorporating uncertainty quantification to increase reliability and decision-making in dynamic variable process environments.