CORMSIS Seminar "Research in the time of Covid-19: novel deterministic and stochastic modelling approaches to epidemiology-based decision support" - Edilson Fernandes De Arruda Event

Event details

The current epidemic highlighted the need for robust decision support tools combining epidemiology, mathematical modelling and operational research. Multiple viral strains, reinfection and waning immunity are some of the challenging issues that became evident as the Covid-19 epidemic unfolded. These issues are separately studied in the literature, but so far a unified framework is still lacking. This talk will focuses on deterministic and stochastic models based on the classical SEIR framework and with distinct levels of generality. We introduce a general deterministic decision support tool that considers reinfection, multiple viral strains and an optimal control formulation designed to cope with the tradeoff between infection levels and the economic consequences of mitigation strategies. We provide insights into the decision making implications of multiple viral strains and on the shape of the resulting optimal control policies in the light of the Covid-19 epidemic in Brazil. We also introduce a novel stochastic approach to SEIR modelling that uses insights from queuing theory to propose a Markov process that emulates the evolution of the disease whilst also considering the underlying uncertainty. In contrast to the previous literature, the model simultaneously considers general latency and recovery periods without loss of computational tractability. We propose a class of stabilising mitigation policies based on the analogy between occupation rates and the reproduction number (R0) and evaluate their effect on a series of validating experiments.