Friday, April 10, 2026 at 12:00 pm

JILA Foothills Room

Mark Rast, CU Boulder

"nverting epidemic trajectories for disease attributes"

Abstract:

Reliable inference of disease properties is required for public health decisions. For novel infectious agents these may not be obtainable on short time scales by understanding the biology of the pathogen. We assess the ability to infer infectious disease attributes from the progression of an outbreak in a population. We construct stochastic Kermack-McKendrick trajectories, sample them with and without reporting error, and evaluate our ability to invert the population behavior for the population mean infectiousness as a function of time since infection and infection duration distribution. We show that, for a well-mixed population, Poisson GLM regression can recover the integral kernels used in the construction of the stochastic epidemics. Either multi-trajectory or regularized single trajectory inversions are effective. Moreover, we demonstrate that the individual infectiousness profile, the infectiousness of an individual over the duration of their infection, can be found using the infection duration distribution and population mean infectiousness recovered, assuming self-similarity of the individual infectious profiles over the infectious period. Our work suggests that aggressive monitoring of the stochastic evolution of a novel infectious disease outbreak in a single local well-mixed population can allow determination of the underlying disease attributes that characterize its spread.

Back to Speakers