Patient digital twins

Patient digital twins combine mechanistic models (growth, radiobiology, pharmacokinetics, immune dynamics) with a patient's own imaging, genomics and markers, re-fitted as new data arrive. Prototypes forecast glioma growth for radiotherapy planning, optimise radiation fractionation for individual patients, simulate CAR-T expansion and choose adaptive therapy thresholds; the US National Academies set out a research agenda for digital twins in 2023 and the FDA has begun to discuss their evidentiary standards. They differ from virtual control arms, which model a typical patient rather than this one, and they stand or fall on the identifiability of their parameters from sparse clinical data.