Evolution of human pharmacokinetics prediction methods: from allometry to artificial intelligence

Accurate prediction of human pharmacokinetics (PK) remains a critical challenge in drug development. Conventional animal-based approaches face inherent limitations due to species differences, driving progress toward more accurate and human-relevant predictive methodologies. This review discusses the evolution of human PK prediction methodologies from conventional approaches to emerging technologies. Traditional methods, including allometric scaling, in vitro–in vivo extrapolation, and physiologically based pharmacokinetic modeling, are first examined, with emphasis on their principles, applications, and limitations through representative case studies. Subsequently, emerging approaches, such as artificial intelligence (AI) and machine learning applications, organ-on-a-chip systems, and other advanced in vitro models, are explored. A comprehensive comparison between conventional and AI-based approaches is provided, addressing data requirements, modeling complexity, predictive performance, and practical applicability. Current implementation challenges—including the need for standardization, barriers to cross-method integration, and regulatory considerations—are also discussed. The future of human PK prediction depends on the strategic integration of complementary methodologies. Conventional methods provide mechanistic transparency, AI-driven approaches offer screening efficiency, and advanced in vitro models deliver physiological relevance. Achieving this integration requires standardized validation frameworks and clear regulatory guidance to support multi-method applications.