Speaker
Description
In the wake of the SARS-CoV-2 pandemic, there has been increased interest in mathematical tools that can anticipate changes in infectious disease dynamics from surveillance data. Early warning signals based on critical slowing down have been widely proposed as indicators of tipping points such as disease elimination or resurgence. However, many epidemiological systems are non-stationary and may exhibit transient increases in fluctuations even in the absence of changes in the basic reproduction number.
In this work, we show how critical-like signals can arise from the geometry of stochastic dynamics when the underlying deterministic system is non-normal. We demonstrate that increases in fluctuation variance and recovery time can occur without a change in asymptotic stability. Using a general stochastic framework for fluctuations around mean-field dynamics, we derive conditions under which such behaviour emerges.
As an illustrative example, we analyse the susceptible–infectious–recovered (SIR) model and show how these signals can precede epidemic waves. Our results suggest that critical-like signals may provide useful indicators of impending infection waves in many infectious disease models.
