A data-driven tool for tracking and predicting the course of COVID-19 epidemic as it evolves

New COVID-19 epicenters have sprung up in Europe and US as the epidemic in China wanes. Many mechanistic models past predictions for China were widely off the mark (1, 2), and still vary widely for the new epicenters, due to uncertain disease characteristics. The epidemic ended in Wuhan, and later in South Korea, with less than 1% of their population infected, much less than that required to achieve "herd immunity". Now as most countries pursue the goal of "suppressed equilibrium", the traditional concept of "herd immunity" in epidemiology needs to be re-examined. Traditional model predictions of large potential impacts serve their purpose in prompting policy decisions on contact suppression and lockdown to combat the spread, and are useful for evaluating various scenarios. After imposition of these measures it is important to turn to statistical models that incorporate real-time information that reflects ongoing policy implementation and degrees of compliance to more realistically track and project the epidemics course. Here we apply such a tool, supported by theory and validated by past data as accurate, to US and Europe. Most countries started with a Reproduction Number of 4 and declined to around 1 at a rate highly dependent on contact-reduction measures.