When plotted in what I call ICC (Incidence versus Cumulative Cases) coordinates, noisy disease data appear to fluctuate about a mean curve with generic properties. In this talk, I will recount the discovery of such universality [1] and describe recent work aimed at elucidating this behavior [2, 3]. In particular, exact results will be provided for the deterministic and stochastic SIR models. In addition, I will explain how identifying trends in the ICC plane can lead to short-term forecasts and illustrate this approach on COVID-19 cases and deaths in the US [4].
This is joint work with Hannh Biegel, Bill Fries, Faryad Sahneh, and Joe Watkins.
[1] J. Lega and H.E. Brown, Data-driven outbreak forecasting with a simple nonlinear growth model, Epidemics 17, 19–26 (2016).
[2] J. Lega, Parameter estimation from ICC curves, Journal of Biological Dynamics 15, 195-212 (2021).
[3] F.D. Sahneh, W. Fries, J.C. Watkins, J. Lega, Epidemics from the Eye of the Pathogen, SIAM J. Appl. Math. 82, 2036-2056 (2022).
[4] H. Biegel & J. Lega, EpiCovDA: a mechanistic COVID-19 forecasting model with data assimilation, arxiv.org/abs/2105.05471 (2021).