ML-Driven Pharmaceutical Cocrystal Technology: Advances in Screening, Property Prediction and Applications

Recently, pharmaceutical cocrystal technology has garnered considerable global attention because of its innovativeness and environmental sustainability. This technology effectively enhances the bioavailability of poorly soluble drugs and optimizes their physicochemical and biological properties. Considering the pivotal role of cocrystal screening in improving drug bioavailability, herein, we develop a comprehensive framework for understanding the definition, screening, key characteristics, and applications of drug cocrystals in contemporary drug development. We discuss a series of innovative and efficient screening methods, such as high-throughput computational screening, artificial intelligence screening, and Raman spectroscopy, focusing particularly on the application of machine learning (ML) algorithms. Such algorithms can analyze large volumes of physicochemical data for virtual cocrystal screening and solubility prediction. Compared with traditional experimental screening methods-such as X-ray diffraction, thermal analysis (TA), and high-performance liquid chromatography (HPLC), ML models can manage high workload, address gaps in screening, and facilitate accurate solubility prediction. By using ML models, researchers can narrow the experimental scope and accelerate the discovery and development of novel drug cocrystals. This approach holds promise for enhancing drug efficacy, reducing adverse reactions, and improving patient compliance. Future studies should integrate experimental and virtual screening methods to enhance the efficiency and accuracy of cocrystal selection.