Black boxes and petri dishes: AI models for antimicrobial pharmacokinetics, drug properties, and resistance

The convergence of artificial intelligence (AI) and antimicrobial pharmacology is creating unprecedented opportunities to accelerate antimicrobial discovery, optimize therapeutic strategies, and reduce the global burden of infectious diseases. A critical determinant of the successful integration of AI into antimicrobial pharmacology is the accurate and interpretable prediction of pharmacokinetic (PK) and pharmacodynamic (PD) properties, including measures such as the minimum inhibitory concentration (MIC), to identify the PK/PD indices and dosing regimens that maximize antimicrobial efficacy while ensuring safety. We review recent developments in AI models for antimicrobial PK/PD and critically examine their performance, interpretability, generalizability, and readiness for clinical translation. We provide a practical framework to help researchers, microbiologists, pharmacologists, and clinicians evaluate the strengths, limitations, and appropriate applications of these emerging approaches. We anticipate that rigorous development, transparent validation, and thoughtful implementation of AI-driven PK/PD models will accelerate antimicrobial discovery, optimize precision dosing, and ultimately improve patient outcomes.