Artificial Intelligence in Drug Delivery and Pharmaceutical Sciences: A Critical Review

Abstract

Pharmaceutical research is changing the world. Scientific advancements over decades have led to new molecular entities; however, the costs of developing them have increased to over 2.6 billion, and only 12% of Phase I chemical entities obtain regulatory approval. The recent emergence of artificial intelligence as a computational paradigm is good because it provides a tool set to address inefficiencies in drug research paradigm. This critical review covers AI applications in drug delivery systems, Quality by Design-based formulation optimization, pharmacokinetic and pharmacodynamic modelling, and clinical trial management. Limitations and drawbacks of Progress are discussed but success stories are highlighted. A narrative literature review was used to address criticism. Searches in Web of Science, PubMed and Scopus. Search terms included ‘artificial intelligence’, ‘machine learning’, ‘deep learning’, ‘drug delivery’, ‘nanoparticles’, ‘pharmacokinetics’, ‘formulation optimisation’ and ‘Quality by Design’. We searched for seminal work prior to 2015 and work from 2015 up to March 2026. Machine learning, deep learning, generative models, and NLP have improved target identification, lead optimization, nanoparticle development, and personalized medicine. AI has simplified the formulation optimization process in Quality by Design and improved the prediction of encapsulation efficiency, size distribution, and in vitro and in vivo release kinetics in nanoparticles, liposomes, and self-emulsifying systems. Pharmaceutical science has advanced with AI not just through faster computation but by solving problems previously unsolvable. This potential can be realized through sustained, multi-disciplinary efforts to improve data quality, model interpretability, and regulator acceptance. At all stages, pharmacy professionals need to have basic AI literacy.