Charles L Howe
Drug development is slow, costly, and prone to late-stage failure, in part because animal models poorly predict human responses. Two human-relevant technologies are maturing in parallel: biological avatars, defined as patient- or stem-cell-derived models such as organoids and organ-on-a-chip systems, and digital twins, defined as computational models that integrate a patient's molecular and clinical data to forecast treatment responses. We propose the ex vivo clinical trial concept, in which an avatar and a digital twin are coupled in an iterative loop so that laboratory measurements refine the computational prediction and the prediction guides the next experiment, allowing candidate therapies to be tested and prioritised before a patient is exposed. We review the platforms, their predictive performance in cancer, cystic fibrosis, and liver toxicity, the conditions under which they fail, and the qualification, turnaround, and standardisation requirements that must be met before such trials can inform drug development or clinical care.