The COVID-19 pandemic demonstrated the value of modeling to inform health policy. Models were used to provide situational awareness and inform mitigation policies. However, uncertainty surrounding the longevity of vaccine and infection-induced immunity, the emergence and characteristics of SARS‐CoV‐2 variant strains, and behavioral responses to policy interventions prevent modelers from providing more than a few weeks of model-based foresight. Under those conditions, policymakers have options to control the pandemic, but deep uncertainties deny the prediction of their long-term effects. Robust Decision Making (RDM) is a set of methods and tools designed to inform decisions under conditions of deep uncertainty. This dissertation presents three papers exploring the utility of RDM for supporting health policy decisions. The first paper discusses how public health decision-makers may benefit from RDM, using pandemic response policies as a motivating example. The second paper presents a stress test of California's COVID-19 reopening strategy, demonstrating that adaptive reopening plans are superior to non-adaptive ones. The third evaluates the robustness of colorectal cancer screening strategies to uncertainties surrounding the natural history of the disease. Finally, this dissertation reflects on the broader applicability of RDM to health policy decision-analytic problems and reflects on future research directions.
Table of Contents
COVID-19 and Deep Uncertainty
Robustness of Colorectal Cancer Screening