Integrating artificial intelligence into drug delivery systems: Formulation development and current challenges

Artificial intelligence (AI) is increasingly being explored as a supportive tool to address persistent challenges in drug delivery research, particularly those associated with formulation complexity, optimization efficiency, and controlled drug release. While AI applications in drug discovery are well established, their role in drug delivery systems remains less structured and is primarily focused on supporting formulation design, optimization, and performance prediction within experimentally constrained environments. Conventional trial-and-error approaches are resource-intensive and often struggle to capture the multidimensional relationships between formulation variables, carrier properties, and biological performance. In this context, AI-driven methods offer opportunities to enhance data interpretation and guide formulation development in a more systematic and predictive manner. This review examines the application of AI across key stages of drug delivery development, including formulation design, nanocarrier optimization, and smart and controlled release systems. Machine learning (ML) and deep learning (DL) approaches applied to pre-formulation analysis, excipient selection, particle engineering, and release-profile optimization are discussed, highlighting their role as decision-support tools rather than autonomous systems. The integration of AI within Quality by Design (QbD) frameworks is also critically assessed, with particular attention to its potential to support the identification of critical formulation variables and design-space exploration. Despite these advances, translation into routine pharmaceutical practice remains limited by challenges related to data quality, experimental validation, model interpretability, and biological complexity. Overall, this review positions AI as a complementary and decision-support layer in drug delivery development and highlights the key barriers that must be addressed for its reliable, scalable, and regulatory-aligned implementation.