Hierarchical Models for Covid - identifying effects of lockdown and an R package

The epidemia package allows researchers to flexibly specify and fit Bayesian epidemiological models in the style of Flaxman et al (Nature 2020) and make inference about the time-varying reproduction number R(t) and the number of infections over time based on death counts. The package leverages R’s formula interface to paramaterize the time-varying reproduction rate as a function of covariates. Multiple populations can be modeled simultaneously with hierarchical models. The design of the package has been inspired by, and has borrowed from, rstanarm (Goodrich et al. 2020). epidemia uses Stan as the backend for fitting models. We will discuss some of the recent controversies surrounding the impact of non-pharmaceutical interventions such as stay-at-home / lockdown orders and the stability of these models (using a multiverse analysis, and prior and posterior predictive checks) when fit to individual countries vs. multiple countries.

Documentation:

https://imperialcollegelondon.github.io/epidemia

https://mrc-ide.github.io/covid19usa

https://github.com/ImperialCollegeLondon/covid19model

Presenter biography:
Seth Flaxman

Seth Flaxman is a senior lecturer in statistical machine learning in the statistics section of the Department of Mathematics at Imperial College London. I help lead the Machine Learning Initiative at Imperial and the EPSRC Centre for Doctoral Training in Modern Statistics and Statistical Machine Learning (StatML) at Imperial and Oxford. My research is on scalable methods and flexible models for spatiotemporal statistics and Bayesian machine learning, applied to public policy and social science. I’ve worked on application areas that include public health, crime, voting patterns, filter bubbles / echo chambers in media, the regulation of machine learning algorithms, and emotion.