Deep learning for small-molecule drug discovery: From molecular design to clinical translation

Recent advances in artificial intelligence (AI) are increasingly transforming drug discovery, offering new approaches to accelerate the identification and optimization of therapeutic candidates. Deep learning (DL) methods have shown strong potential for generative modeling and molecular property prediction, enabling the design of compounds with tailored pharmacological profiles. In this review, we synthesize recent AI-driven progress in drug discovery, highlighting both achievements and ongoing challenges. We discuss key technical aspects underpinning DL-based models, including the use of curated molecular databases, molecular descriptors, and standardized evaluation strategies. We examine representative architectures and illustrate how they have been applied to molecular generation, binding affinity prediction, and multi-modal integration of ligand and protein data. In addition, we provide a comprehensive analysis of AI-enabled small-molecule drugs, including discovered and repurposed molecules in clinical trials, and explore the intellectual property information and chemical structures of these compounds. We demonstrate that AI-native companies are evolving pharmaceutical pipelines by integrating these tools at various stages of development. We also address the regulatory framework of agencies such as the United States Food and Drug Administration (FDA) and the European Medicines Agency (EMA) as they adapt to integrate AI innovations. Finally, we outline persistent limitations, such as data inconsistency, model interpretability, and the gap between benchmark performance and real-world applicability, which remain critical barriers to widespread adoption. By consolidating technical advances and open challenges, this review aims to provide a balanced perspective on the role of AI in reshaping modern drug discovery. • Overview of cutting-edge AI models for molecular design and property prediction. • Presentation of key data sources, descriptors, and evaluation benchmarks. • Description of the evolution of models for molecular property prediction. • Enumeration of promising AI-generated small molecules progressing in clinical trials. • Survey of the regulatory framework responses to AI-driven drug development tools.