Artificial Intelligence-Driven Drug Discovery and Development: From Molecular Design to Clinical Translation

Abstract

Artificial Intelligence (AI) is increasingly reshaping pharmaceutical research by enabling computational approaches that can accelerate the identification, optimization, and development of therapeutic candidates. Conventional drug discovery is characterized by substantial financial investment, lengthy development timelines, and high attrition rates, particularly during clinical evaluation. AI-driven methods offer opportunities to address several of these limitations through machine learning, deep learning, generative models, molecular docking, virtual screening, and predictive pharmacology. This review examines the evolving role of AI across the drug discovery and development continuum, beginning with target identification and molecular design and extending to preclinical evaluation and clinical translation. Particular attention is given to structure-based and ligand-based drug design, prediction of pharmacokinetic and toxicological properties, de novo molecular generation, biomarker discovery, patient stratification, and clinical trial optimization. The review also considers how artificial intelligence can integrate heterogeneous biological, chemical, genomic, and clinical datasets to support more informed therapeutic decisions. A focused literature synthesis was undertaken using published studies addressing AI applications in pharmaceutical research and development. The available evidence indicates that AI can improve virtual screening efficiency, prioritize promising molecular candidates, identify previously unrecognized biological relationships, and support optimization of drug-like properties. However, successful clinical translation remains dependent on data quality, model interpretability, external validation, regulatory oversight, reproducibility, and appropriate integration with experimental research. Important challenges include dataset bias, limited representation of biological complexity, inconsistent benchmarking, privacy concerns, and the risk of overestimating computational predictions. Overall, AI should be viewed not as a replacement for experimental pharmacology but as a complementary technology that can strengthen decision-making throughout drug development. Continued collaboration among computational scientists, medicinal chemists, pharmacologists, clinicians, and regulatory authorities will be essential for translating AI-generated hypotheses into safe, effective, and clinically meaningful therapeutics.