In the aftermath of a sudden catastrophe, first responders (FRs) strive to reach and rescue immobile victims. Simultaneously, civilians use the same roads to evacuate, access medical facilities and shelters, or reunite with their relatives via private vehicles. The escalated traffic congestion can significantly hinder critical FR operations. A proposal from the Türkiye Ministry of Transportation and Infrastructure is to allocate a lane on specific road segments exclusively for FR use, mark them clearly, and precommunicate them publicly. For a successful implementation of this proposal, an FR path should exist from designated entry points to each FR demand point in the network. The reserved FR lanes along these paths will be inaccessible to evacuees, potentially increasing evacuation times. Hence, in this study, we aim to determine a subset of links along which an FR lane should be reserved and analyze the resulting evacuation flow under evacuees’ selfish routing behavior. We introduce this problem as the first responder network design problem (FRNDP) and formulate it as a mixed-integer nonlinear program. To efficiently solve FRNDP, we introduce a novel bilevel nested heuristic, the Graver augmented multiseed algorithm (GAMA) within GAMA, called GAGA. We test GAGA on synthetic graph instances of various sizes as well as scenarios related to a potential Istanbul earthquake. Our comparisons with a state-of-the-art exact algorithm for network design problems demonstrate that GAGA offers a promising alternative approach and highlights the need for further exploration of quantum-inspired computing to tackle complex real-world problems. History: Accepted by Giacomo Nannicini, Area Editor for Quantum Computing and Operations Research. Accepted for Special Issue. Funding: S. Tayur and A. Tenneti acknowledge Raytheon BBN (RTX-BBN) for its support through a Carnegie Mellon University-BBN contract as part of a Defense Advanced Research Projects Agency project on quantum-inspired classical computing. A. Karahalios is supported by the National Science Foundation Graduate Research Fellowship Program [Grants DGE1745016, DGE2140739]. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2024.0574 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2024.0574 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .

Given only data generated by a standard confounding graph with unobserved confounder, the Average Treatment Effect (ATE) is not identifiable. To estimate the ATE, a practitioner must then either (a) collect deconfounded data; (b) run a clinical trial; or (c) elucidate further properties of the causal graph that might render the ATE identifiable. In this paper, we consider the benefit of incorporating a large confounded observational dataset (confounder unobserved) alongside a small deconfounded observational dataset (confounder revealed) when estimating the ATE. Our theoretical results suggest that the inclusion of confounded data can significantly reduce the quantity of deconfounded data required to estimate the ATE to within a desired accuracy level. Moreover, in some cases—say, genetics—we could imagine retrospectively selecting samples to deconfound. We demonstrate that by actively selecting these samples based upon the (already observed) treatment and outcome, we can reduce sample complexity further. Our theoretical and empirical results establish that the worst-case relative performance of our approach (vs. a natural benchmark) is bounded while our best-case gains are unbounded. Finally, we demonstrate the benefits of selective deconfounding using a large real-world dataset related to genetic mutation in cancer.