ABSTRACT
Policy makers, researchers and public health officials rely on forecasts from infectious disease models to
anticipate resource use, assess risk, and plan public health interventions. In order to effectively and accurately
predict future cases and hospitalization rates of COVID-19, these models depend on high quality data on
infections in the population. However, as clinical testing becomes less reliable with the ubiquity of at-home
antigen tests and the frequency of case reporting is decreased in many states, there is a critical need to identify
alternative data sources. Wastewater-based surveillance is a promising supplement to traditional data sources
for monitoring infection at the population level: it is anonymous, cost-effective, and unaffected by variation in
clinical testing and reporting. It also reflects infections in the population days or weeks sooner than they would
be captured by reported cases or hospitalizations. While there is ample evidence that wastewater concentrations
of SARS-CoV-2 correlate with COVID-19 cases, wastewater data have rarely been used to inform modeling and
forecasting, and the benefit of incorporating wastewater surveillance data into data-driven dynamical models for
forecasting has not been rigorously assessed. The goal of this project is to leverage a rich dataset from the New
York State Wastewater Surveillance Network to evaluate the potential for wastewater surveillance data to
improve inference of SARS-CoV-2 transmission dynamics. Partnering with the New York State Wastewater
Surveillance Network, which includes over 150 wastewater treatment plants across 61 counties of New York
State and represents a population of more than 15 million, provides a unique opportunity to investigate the value
of wastewater surveillance for modeling at a larger scale. We will build on an existing metapopulation model
developed by the Shaman Lab to test how inference and forecasting of transmission dynamics can be improved
by integrating wastewater data. In other words, we will use retrospective data from New York State to rigorously
assess: can we predict future cases more accurately by using wastewater data to inform our models? This study
would be the first that we know of to use a metapopulation structure, which considers population movement, to
model SARS-CoV-2 with wastewater surveillance data. By more realistically modeling how human populations
move and mix, this model structure can better capture real-world transmission dynamics. This will be crucial to
realistically evaluate whether wastewater data can be relied on in the future to forecast COVID-19 incidence,
which will be particularly important as clinical testing fluctuates and case reporting becomes sparser. It may also
motivate the application of wastewater-based modeling to other pathogens detectable in feces, facilitating the
development of prompt, cost-effective forecasts of other communicable diseases in the future.
