Causal Modeling of Organ Offer Dynamics for Better Allocation

Organ transplantation remains severely constrained by donor scarcity and in-
efficiencies in offer acceptance. Across organ types, acceptance rates in real-world
systems commonly fall below 5%, with most offers refused multiple times before
a successful placement [3]. These refusals prolong cold ischemic time (CIT), de-
grade graft quality, and substantially increase the probability of eventual discard
[4]. Any allocation policy aiming to improve health outcomes must therefore
model not only long-term transplant benefit, but also acceptance behavior, de-
cision time, and the sequential dynamics governing organ deterioration. Yet
current allocation models, including those embedded in regulatory simulators,
rely on oversimplified, non-causal acceptance estimators, failing to generalize to
counterfactual policies or to incorporate clinician-provided explanatory signals.