Optimal Routing for Mobile Lung Cancer Screening Vehicle Guided by Incidence Rates: A Case Study in Oklahoma

Abstract

Oklahoma ranks 8th in the United States for age-adjusted lung cancer incidence and 5th for lung cancer mortality, making it one of the most severely affected regions in the country. Large randomized clinical trials including the National Lung Screening Trial (NLST) have demonstrated that low-dose computed tomography (LDCT) screening reduces lung cancer mortality by approximately 20% among high-risk individuals compared with chest X-ray, leading the U.S. Preventive Services Task Force to issue a Grade B recommendation for LDCT screening. Despite this strong evidence base, screening uptake remains persistently low nationwide (approximately 16%) and is even lower in Oklahoma, where 11% of eligible individuals are screened each year. This gap is driven in part by limited geographic access: only a minority of Oklahoma counties have LDCT facilities, and rural residents face significant transportation and socioeconomic barriers.In response, multiple healthcare systems have initiated mobile LDCT screening programs to expand access, particularly in underserved and medically isolated communities. However, most existing mobile programs are deployed reactively, following ad hoc or invitation-based scheduling. This approach fails to strategically allocate resources, does not account for geographic and epidemiological disparities, and lacks a systematic framework to maximize health benefits relative to operational costs. To support a transition from reactive deployment to a strategic, data-driven model, this study develops a two-stage stochastic routing optimization framework for mobile lung cancer screening under uncertainty in community participation. The proposed model seeks to maximize expected quality-adjusted life years (QALYs) gained from early detection while incorporating operational constraints, travel distances, screening capacity, financial considerations, and scenario-based uncertainty in county-level uptake. The model is parameterized using county-stratified incidence data from the Oklahoma Central Cancer Registry, county demographic and geospatial characteristics, and evidence-based estimates of screening benefits, false-positive harms, and QALY outcomes. We apply the model to all Oklahoma counties lacking LDCT facilities and evaluate alternative deployment strategies under multiple willingness-to-pay thresholds and QALY disutility assumptions. Results demonstrate how optimized routing can expand access, improve early detection in high-need regions, and quantify the levels of societal incentive required to sustain mobile screening economically. This work provides a rigorous, evidence-based foundation for designing efficient, equitable, and proactive mobile lung cancer screening programs in Oklahoma and offers a generalizable framework for other states facing similar disparities in cancer screening access.