Real-time Dissection and Forecast of Infection Dynamics during a Pandemic

Pandemic preparedness requires institutions, including public health authorities and governments, to detect, survey and control outbreaks. To maintain an accurate, quantitative and up-to-date picture of an epidemic crisis is key. For SARS-CoV-2, this was mostly achieved by ascertaining incidence numbers and the effective reproductive number ( R eff ), which counts how many people an infected person is likely to infect on average. These numbers give strong hints on past infection dynamics in a population but fail to clearly characterize current and future dynamics as well as potential effects of pharmaceutical and non-pharmaceutical interventions. We show that, by using and combining infection surveillance and population-scale contact statistics, we can obtain a better understanding of the drivers of epidemic waves and the effectiveness of interventions. This approach can provide a real-time picture, thus saving not only many lives by quickly allowing adaptation of the health policies but also alleviating economic and other burdens if an intervention proves ineffective. We factorize R eff into contacts and relative transmissibility: Both signals can be used, individually and combined, to identify driving forces of an epidemic, monitoring and assessing interventions, as well as projecting an epidemic’s future trajectory. Using data for SARS-CoV-2 and Influenza from 2019 onward in Germany, we provide evidence for the usefulness of our approach. In particular, we find that the effects from physical distancing and lockdowns as well as vaccination campaigns are dominant.