AI AUGUMENTED PBPK MODELLING: TRANSFORMING PHARMACOKINETIC PREDICTION AND DRUG DEVELOPMENT

Hari Harshanth*, Shailaja Pashikanti, Mylapalli Durga Devi

Physiologically based pharmacokinetic (PBPK) modeling is widely used to predict drug absorption, distribution, metabolism, and excretion using mechanistic representations of human physiology. PBPK models support informed decision-making in drug development, regulatory evaluation, and risk assessment; however, their application is often limited by extensive data requirements, parameter uncertainty, and challenges in estimating drug-specific inputs, particularly at early stages of development. Artificial intelligence (AI) and machine learning (ML) have recently emerged as valuable complementary tools that can address these limitations. By learning complex, non-linear relationships from molecular descriptors, in vitro data, and simulation outputs, AI-based models enable rapid prediction of key pharmacokinetic parameters. The integration of AI with PBPK modeling has led to hybrid AI–PBPK frameworks that combine the physiological interpretability of mechanistic models with the predictive efficiency and scalability of data-driven approaches. This review provides an overview of AI-driven approaches in PBPK modeling, covering fundamental PBPK principles, commonly used AI and ML techniques, and strategies for integrating machine learning outputs into PBPK simulations. Key applications in drug development, including pharmacokinetic prediction, drug–drug interaction assessment, formulation optimization, and special population analysis, are discussed. The review also highlights current challenges related to data quality, model interpretability, and regulatory acceptance, and outlines future perspectives for advancing AI-assisted PBPK modeling.