Design of alliance-based coordination mechanisms for renal relief

Abstract

Disasters often render hemodialysis clinics inoperable, posing severe risks to patients with end-stage kidney disease (ESKD), who depend on regular dialysis for survival. When services are disrupted, patients must be transferred rapidly to operational clinics. Establishing a coordination mechanism before a disaster can support timely matching of patients to available clinics and reduce post-disaster chaos. We introduce the Alliance Network Design Problem, which proactively forms alliances among clinics to support coordinated renal relief. These alliances enable resource pooling and pre-assignment of backup clinics to patients. We formulate the ANDP as a two-stage stochastic mixed integer programming model that determines alliance relationships and backup clinic assignments under uncertainty in post-disaster clinic capacities. To address computational challenges, we propose a Hybrid-Sample Average Approximation (H-SAA) method integrating Latin Hypercube Sampling and Importance Sampling. We show that H-SAA outperforms conventional SAA approaches, particularly by better capturing low-probability, high-impact scenarios. A case study of the hemodialysis network in Istanbul, Türkiye, demonstrates the effectiveness of the proposed approach in strengthening disaster preparedness for ESKD patients. Our results show that alliance formation mitigates the adverse effects of disasters on dialysis continuity relative to a no-alliance setting. We also provide policy-relevant insights by evaluating complementary resilience strategies, including region-specific clinic fortification and post-disaster capacity augmentation.