Beyond Three Tiers: Subtype-Aware Stochastic Optimization and Resource Misallocation Measurement for Earthquake Emergency Medical Response

Optimizing emergency medical operations after major earthquakes requires coordinating temporary facility deployment, patient routing, and staff allocation under deep uncertainty, within the first 72 hours when survival chances are highest. Standard triage-based optimization models lump all casualties into three tiers, ignoring clinically meaningful variation in treatment needs that translates directly into misallocated resources. We present three contributions applicable to any triage-aware stochastic model in this class: (i) an eight-subtype casualty classification (T1a– T1c, T2a–T2c, T3a–T3b) with subtype-specific Markov survival dynamics; (ii) Monte Carlo scenario generation calibrated to Istanbul earthquake loss data, replacing deterministic linear scaling; and (iii) the Resource Misallocation Index (RMI), a practitioner-ready metric converting classification coarseness into wasted doctor-hours. Experiments on Istanbul’s Kartal district show the three-tier baseline misallocates 22.1 % of all doctor-minutes (131 doctor-hours per event); 696,550- variable instances solve in 69 seconds; and 6–10 additional lives are saved under 3×–5× capacitystressed scenarios. Results hold under ±40 % subtype-proportion perturbation and up to 30 % triage error.