Abstract — In this work we focus on large-scale infectious disease screening in dynamic and highly variable environments. Large-scale screening is crucial to any well-functioning healthcare system as it has direct impact on human life as was demonstrated in the ongoing COVID -19 pandemic. Problems that arise in this context are often sequential as they involve taking screening actions, observing outcomes (which are often affected by uncertainty), and then adjusting decision accordingly. Given the nature of the problems involved, decision that are more conservative (as opposed to ones that rely on average-case performance) are desirable. Therefore, we rely on tools such as robust dynamic programming and optimal control to solve the resulting sequential decision-making problems. We focus on two aspects of the problem. The first deals with large-scale screening of established infections in which disease evolution can be predicted with a fairly good level of accuracy. In this setting, screening policies do not have a substantial impact on disease prevalence in the short-to-medium term and thus interactions between screening and populations characteristics can be relaxed. The second aspect focuses on large-scale screening for emerging pandemics. In this setting, modeling the interaction between screening decisions and disease dynamics on the short-term is crucial and thus must be explicitly incorporated in the decision-making framework. This unfortunately complicates the problem from a modeling perspective. For both problem aspects, we propose efficient algorithms and present case studies in the context of large-scale screening of donated blood in blood centers and of COVID-19 in the U.S.