CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Incorporating infectious duration-dependent transmission into Bayesian epidemic models Authors:  Caitlin Ward - University of Calgary (Canada) [presenting]
Abstract: Compartmental models are commonly used to describe the spread of infectious diseases by estimating the probabilities of transitions between important disease states. A significant challenge in fitting Bayesian compartmental models lies in the need to estimate the duration of the infectious period, based on limited data providing only symptom onset date or another proxy for the start of infectiousness. Commonly, the exponential distribution is used to describe the infectious duration, an overly simplistic approach that is not biologically plausible. More flexible distributions can be used, but parameter identifiability and computational cost can worsen for moderately-sized or large epidemics. We present a novel approach that considers a curve of transmissibility over a fixed infectious duration. Incorporating infectious duration-dependent (IDD) transmissibility, which decays to zero over the infectious period, is biologically reasonable for many viral infections. Fixing the maximum length of the infectious period eases computational complexity in model fitting. Through simulation, we evaluate different functional forms of IDD transmissibility curves and show that the proposed approach offers an improved estimation of the time-varying reproductive number. The benefit of our approach is illustrated through a new analysis of the 1995 outbreak of Ebola Virus Disease in the Democratic Republic of the Congo.