University of Cambridge > > Machine Learning @ CUED > Estimating RSV seasonality from pandemic disruptions: a modelling study

Estimating RSV seasonality from pandemic disruptions: a modelling study

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Hong Ge.

Background: Respiratory syncytial virus (RSV) is a leading cause of respiratory tract infections and bronchiolitis in young children. The seasonal pattern of RSV is shaped by short-lived immunity, seasonally varying contact rates and pathogen viability. The magnitude of each of these parameters is not fully clear. The disruption of the regular seasonality of RSV during the COVID pandemic in 2020 due to control measures, and the ensuing delayed surge in RSV cases provides an opportunity to disentangle these factors and to understand the implication for vaccination strategies. A better understanding of the drivers of RSV seasonality is key for developing future vaccination strategies. Methods: We developed a mathematical model of RSV transmission, which simulates the sequential re-infection (SEIRRS4) and uses a flexible Von Mises function to model the seasonal forcing. Using MCMC we fit the model to laboratory confirmed RSV data from 2010-2022 from NSW while accounting for the reduced contact rates during the pandemic with Google mobility data. We estimated the baseline transmission rate, its amplitude and shape during RSV season as well as the duration of immunity. The resulting parameter estimates were compared to a fit to pre-pandemic data only, and to a fit with a cosine forcing function. We then simulated the expected shifts in peak timing and amplitude under two vaccination strategies: continuous and seasonal vaccination. Results: We estimate that RSV dynamics in NSW can be best explained by a high effective baseline transmission rate (2.94/d, 95% CrI 2.73-3.18) and a narrow peak with a maximum 13% increase compared to the baseline transmission rate. We also estimate the duration of post infection temporary but sterilizing immunity to be 412 days (95% CrI 391-435). Including data from the pandemic period in the fit reduced parameter correlation substantially and improved parameter identifiability. The continuous vaccination strategy led to more extreme seasonal incidence with a delay in the peak timing and a higher amplitude whereas seasonal vaccination flattened the incidence curves.

This talk is part of the Machine Learning @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity