- This event has passed.
Greenberg Lectures: Regression Models for COVID-19 Epidemic Dynamics with Incomplete Data
The 2021 Bernard G. Greenberg Distinguished Lecture Series
Featuring Professor Xihong Lin, Harvard University
Lecture 3: May 21, 10-11 a.m.
Regression Models for COVID-19 Epidemic Dynamics with Incomplete Data
Modeling infectious disease dynamics has been critical throughout the COVID-19 pandemic. Of particular interest are the incidence, total prevalence, and effective reproductive number (Rt). Estimating these quantities is challenging due to under-ascertainment, unreliable reporting, and time lags between infection, onset, and testing. Dr. Lin proposes a Multilevel Epidemic Regression Model to Account for Incomplete Data (MERMAID) to jointly estimate Rt and ascertainment rates, and thereby predict incidence and total prevalence, over time in one or multiple regions. To account for under-ascertainment, the research team (a) modeled the ascertainment probability over time as a function of testing metrics and (b) jointly model data on confirmed cases and population-based serological surveys. To account for delays between infection, onset, and reporting, they modeled stochastic lag times as missing data. To model Rt over time in one or multiple regions, they specified a semi-parametric function that can incorporate geographic and time-varying covariates. They evaluated MERMAID in simulation studies and applied it to analyze COVID-19 daily case counts, PCR testing data and serological survey data across the United States. They found that U.S. states with greater proportions of ascertained cases tend to have had earlier outbreaks and greater total prevalence. They also found substantial differences in Rt associated with specific state containment policies, such as face-covering mandates and gathering restrictions.
