Progress of Machine Learning in Molecular Crystal Design and Crystallization Development

Machine learning (ML) can optimize the research paradigm and shorten the time from discovery to application of novel functional materials, pharmaceuticals, and fine chemicals. Besides supporting material and drug design, ML is a potentially valuable tool for predictive modeling and process optimization. Herein, we first review the recent progress in data-driven ML for molecular crystal design, including property and structure predictions. ML can accelerate the development of the solvates, co-crystals, and colloidal nanocrystals, and improve the efficiency of crystal design. Next, this review summarizes ML algorithms for crystallization behavior prediction and process regulation. ML models support drug solubility prediction, particle agglomeration prediction, and spherical crystal design. ML-based in situ image processing can extract particle information and recognize crystal products. The application scenarios of ML algorithms utilized in crystallization processes and two control strategies based on supersaturation regulation and image processing are also presented. Finally, emerging techniques and the outlook of ML in drug molecular design and industrial crystallization processes are outlined.