Allocation region design for the us liver transplantation network: Simulation-based optimization approach with motivated metamodeling.

摘要

End-Stage Liver Disease (ESLD) is the 12th leading cause of death in the US, accounting for nearly 27,500 deaths in 2006. For most ESLD patients, liver transplantation is the only viable therapy. United Network for Organ Sharing (UNOS) is responsible for allocating cadaveric livers in the US. The current UNOS allocation system is a hierarchical system that prioritizes patients according to two major criteria, geographical proximity of the patient to the donor and the medical urgency of the patient, and allocates the livers according to the priority sequence. The geographical distribution of the livers has been controversial for decades and it remains one of the most pressing concerns in liver allocation at present.;In this research, we develop a simulation-optimization approach that optimizes the allocation region design to maximize quality-adjusted transplant efficiency and geographically-based equity. Our stochastic optimal clustering problem is based on two realistic objectives that cannot be expressed analytically. Since it is time-consuming to evaluate the two objectives for a large number of regional configurations we design motivated surrogates and integrate them with an artificial neural network to develop a meta-model for the stochastic optimization problem. We embed the motivated meta-model in a simulation based meta-heuristic algorithm and show computational results. The new regional configuration obtained shows an improvement from the current regional configuration in terms of allocation efficiency defined by the average cold ischemia time.