Machine-learning-enabled modeling of pharmacokinetics and pharmacodynamics

Computational modeling has been proven to be essential for assessing preclinical and clinical pharmacokinetics (PK) and pharmacodynamics (PD). In recent years, the pharmacometrics community has shown increasing interest in including machine learning (ML) and artificial intelligence (AI) in its computational toolbox, with many possible benefits: more-rapid and -accurate pharmacology assessments, reduced research and development costs, and improved patient safety. As applications increase and PK/PD datasets become more accessible, it is important to stratify the existing AI/ML approaches for PK/PD analysis and position them in the broader context of scientific computing. To that end, this review examines how AI/ML can contribute to PK/PD time-course modeling across various levels, depending on the available data and the question of interest.