Title

Informing Global Polio Eradication with an Integrated Operations Research Modeling Framework

Date

June 24, 2026

Time

1:00 PM – 3:00 PM EST

Format

In-person: Groseclose 402

Online: https://teams.microsoft.com/meet/293288392873511?p=9MHkA2eP5JkxcgeDoc

·                Meeting ID: 293 288 392 873 511

·                Passcode: Nb7UA9wt

Candidate

Yuming Sun

Ph.D. Candidate in Operations Research

H. Milton Stewart School of Industrial and Systems Engineering 

Georgia Institute of Technology

Thesis Committee

• Dr. Pinar Keskinocak (co-advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

• Dr. Lauren N. Steimle (co-advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

• Dr. Yao Xie, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

• Dr. Gian-Gabriel Garcia, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

• Dr. Stephanie D. Kovacs, Global Immunization Division, Centers for Disease Control and Prevention

Abstract

Despite advances in biological and medical science, infectious disease control and prevention continue to face persistent challenges arising from delayed detection, heterogeneous population immunity, and limited resources. These challenges are particularly acute for pathogens that spread largely asymptomatically, such as poliovirus, for which transmission may remain undetected until outbreaks are well established. Effective disease control requires coordinated strategies that integrate disease transmission, surveillance for early detection, and vaccination for outbreak mitigation.

This dissertation develops an integrated operations research modeling framework that links these components through compartmental simulation models, with applications to global polio eradication. The framework consists of three interacting modules: a transmission module that characterizes spatiotemporal disease spread, a vaccination module that evaluates intervention strategies under resource constraints, and a surveillance module that captures imperfect outbreak detection. Together, these modules provide a unified platform for evaluating strategies to predict, detect, and respond to polio outbreaks.

The first part of the dissertation focuses on modeling poliovirus transmission in large, heterogeneous populations. The transmission module extends a classical compartmental simulation model by incorporating heterogeneous immunity resulting from vaccination and infection events, co-circulating virus strains arising from viral mutation, age structure, and geographic heterogeneity. The model is calibrated and validated against confirmed paralytic polio cases using time-series cross-validation that preserves the temporal dependence of transmission dynamics. In a case study of poliovirus transmission in Nigeria, the validated model predicts persistent circulation in under-vaccinated areas and re-emergence in others through spatial exportation. The results suggest that the national vaccination plan may be insufficient to interrupt transmission and highlight the need for proactive preparedness for additional outbreak response activities.

Building on this transmission module, the second and third parts of the dissertation investigate how limited vaccine stockpiles should be allocated during outbreak response. In both studies, the vaccination module completely observes the progression of transmission and determines outbreak response strategies consisting of multiple vaccination campaigns that vary in timeliness, coverage, target populations, geographic scope, vaccine type, campaign duration, and vaccine allocation.

The second part examines the tradeoff between timeliness and coverage while evaluating vaccine allocation based on true immunity. Assuming individual immunity is observable, the model demonstrates the substantial benefits of prioritizing low-immunity children, defined as those lacking sufficient protection against poliovirus infection, compared with the current strategy of vaccinating all eligible children equally. Prioritizing low-immunity children consistently reduces infections and improves outbreak control, even under less favorable levels of timeliness and coverage. The results further demonstrate that, under this targeted allocation strategy, improving timeliness generally yields greater benefits than increasing coverage.

The third part addresses the more realistic setting in which true immunity is not directly observable. Instead, vaccination history is used as a proxy for immunity, and the performance of prioritizing under-vaccinated children is compared with that of prioritizing low-immunity children. The results reveal the consistent superiority of immunity-based allocation, as vaccination history does not fully capture other important determinants of immunity such as prior infection. Nevertheless, prioritizing under-vaccinated children still outperforms the current allocation strategy by reducing infections while using fewer vaccine doses, highlighting the operational value of vaccination history as a practical decision-making tool in resource-limited settings.

The fourth part relaxes the assumption of complete observability of disease transmission by incorporating a surveillance module that captures partially observable transmission dynamics. Multiple surveillance mechanisms are evaluated, including clinical surveillance that detects symptomatic infections, and environmental surveillance that monitors asymptomatic transmission. The interactions among transmission, surveillance, and vaccination are explicitly examined. Results show that improved surveillance sensitivity consistently enables earlier outbreak detection, triggers more timely responses, and reduces disease transmission. As asymptomatic transmission becomes increasingly important in highly vaccinated populations, environmental surveillance emerges as a critical complement to clinical surveillance. Furthermore, expanding environmental surveillance to previously uncovered areas provides greater benefits than intensifying surveillance in already covered areas or uniformly improving clinical surveillance.

The fifth part concludes the dissertation with a critical review of polio modeling studies conducted over the past 25 years. This review highlights successful examples in which modeling informed public health decision-making, identifies simplifying assumptions that have limited the practical implementation of modeling results, and discusses instances in which policy actions diverged from model-based recommendations. It also summarizes key knowledge gaps and outlines priorities for future methodological and applied research.

Overall, this dissertation demonstrates how an integrated operations research modeling framework can support evidence-based polio eradication strategies under real-world constraints. By linking transmission, vaccination, and surveillance, the framework generates actionable insights for policymakers and contributes analytical tools that promote efficient, timely, and equitable strategies for infectious disease control and prevention.