Interpreting antiviral effects: from experimental measurements to mechanistic viral dynamics

Antiviral activity is commonly summarized using endpoint measures such as EC50 or EC90, yet these quantities emerge from a dynamic biological system and depend on viral replication kinetics, host-cell properties, experimental design, and the assay used to quantify infection. Infectious-virus titers, viral RNA, infected-cell fractions, viability measurements, and immune readouts capture different components of the infection process and are therefore not interchangeable measures of "viral load." Mechanistic viral-dynamics models provide a framework for linking such observations to underlying processes, including cell infection, intracellular replication, virus production, cell loss, and host-mediated control. However, the biological information that can be recovered from these models is fundamentally constrained by experimental observability and parameter identifiability. This limitation becomes increasingly important across biological scales: in vitro systems permit dense multimodal measurements and controlled perturbations, whereas animal and human studies provide greater physiological relevance but substantially sparser and more indirect observations. Here, we review how experimental readouts map onto viral-dynamics models, how model structure and parameter inference affect mechanistic interpretation, and how controlled in vitro systems can be used to distinguish competing mechanisms of infection and antiviral action. We further discuss examples in which viral-dynamics modeling has contributed to antiviral pharmacology and regimen selection. Together, these considerations support a view of antiviral effects as context-dependent dynamic phenotypes that require joint interpretation of measurements, models, and experimental design.