(42a) Fast-Tracking Process Optimization with Small Data and Machine Learning

Pharmaceutical process design and optimization is a time-consuming, expensive, and complex task. Data-driven machine learning (ML) can accelerate the process by predicting behavioral trends, variable effects, and optimal conditions to enhance process and product understanding. However, conventional ML algorithms often require extensive datasets, posing a resource-intensive challenge in experimental data collection.
SuntheticsML introduces an enhanced Bayesian optimization (BO) approach with a reaction-agnostic pathway for designing and implementing intelligent experimental campaigns. The insights gained from BO-driven campaigns are used for advanced data analysis and modeling of complex behavioral trends to identify parameter importance, effects, competitive outcomes, and interactions. SuntheticsML is an accessible online ML platform tailored for researchers without coding or ML expertise.

SuntheticsML is a versatile technology that allows numeric, discrete, and mixed-integer optimization problems with up to 20 input and 20 output parameters, further facilitating bounded-target, multi-objective, parametrized, and constrained-input optimizations. Case studies of SuntheticsML with industrial partners showcase accelerated formulation optimization, process characterizations, and process development efforts in chemocatalytic reactions, biocatalytic cascades, crystallizations, in-vitro mRNA transcription processes, and more.

The in-lab validation of SuntheticsML convincingly demonstrates impressive returns on material efficiency, with up to a 75% reduction in the use of expensive or complex reagents. Experimental efficiencies enabling 9-12% increases in previously optimized yields with a 2-32X reduction in optimization experiments.

The insights gained from this work redefine the landscape of reaction engineering, process development, and optimization while simultaneously lowering barriers to the adoption of new chemical technologies.