Groupe d’experts en épidémiologie et en santé publique du Centre de recherches mathématiques

29 octobre 2020 Réunion Zoom

A novel predictive framework for the novel coronavirus

Conférence par Niayesh Afshordi (University of Waterloo, Perimeter Institue for Theoretical Physics)

Abstract: It would be hard to exaggerate the impact that COVID-19 pandemic outbreak, caused by the rapid spread of the SARS-CoV-2 coronavirus, has had on human civilization. Cascading effects from the impact of the pandemic on national healthcare systems, as well as the shutdown of a large fraction of global socioeconomic activity can further impact the health and livelihood of the world population and lead to secondary fatalities, as well as shortening and/or deterioration of lives. Therefore, it is of paramount importance to understand the true dynamics and efficiency of mitigation strategies, so that a proper, transparent, and balanced response can be designed and adopted by local governments across the world. We offer a new data-driven way of attacking this problem via a dynamical causal model informed by our unusual array of backgrounds in cosmology, quantum mechanics, and mathematical modeling. We have developed a physical model for the growth of the disease based on the collision of infected and susceptible populations in a community, with a cross-section+stochastic incubation of the virus. We then calibrate the cross-section and incubation, in terms of population demographics of the county, its Google social mobility, search trends, and weather, by comparing the model to the actual growth rates of COVID in all US counties. This leads to a powerful model (with fewer than a dozen parameters) that can be used to predict the growth/decay of pandemic across the United States, through our online dashboard: . We hope this framework can be scaled up to more regions/data, and be used to inform smart region-specific policies to suppress and/or mitigate the pandemic and its adverse effects.