Estimating the epidemic reproduction number from temporally aggregated incidence data: a statistical modelling approach and software tool

Rebecca K Nash¹, Samir Bhatt^1,2, Anne Cori¹, Pierre Nouvellet^1,3

¹ MRC Centre for Global Infectious Disease Analysis, Imperial College London, UK
² Section of Epidemiology, Department of Public Health, University of Copenhagen
³ School of Life Sciences, University of Sussex, UK

This repository contains all code necessary to reproduce the analysis and figures as described in Nash et al 2023.

Software dependencies

You will need the latest version of EpiEstim to run the analysis. Please install using the following:

install.packages('EpiEstim', repos = c('https://mrc-ide.r-universe.dev',
'https://cloud.r-project.org'))

How to run

setwd("~/EM_EpiEstim_Nash2023")

Data used

The 'real_data' folder contains all of the real-world data (US influenza and UK COVID-19) used in the main paper.

The 'supplementary_analysis' folder contains the simulated data used in the supplementary analysis.

Code

Within the 'main_analysis' folder you will find all code necessary to recreate the influenza and COVID-19 figures in the main paper:

• flu.R
• covid_cases.R
• covid_deaths.R

Within the 'supplementary_analysis' folder you will find all code necessary to recreate the analysis in the appendix.

Supplementary analysis for real-world data:

• flu_appendix.R
• covid_cases_appendix.R
• covid_deaths_appendix.R
• incidence_weekday.R

Simulation study:

• constant_rt.R
• time_varying_rt_sudden.R
• time_varying_rt_gradual.R
• weekend_effects.R
• different_aggregations_aligned.R
• different_aggregations_misaligned.R
• mid_aggregation_variations.R
• loess_smoothing.R

Vignettes

For other worked examples, FAQs, and more details about how to apply EpiEstim to temporally aggregated data, please see the following vignette available in the EpiEstim R package:

• https://mrc-ide.github.io/EpiEstim/articles/EpiEstim_aggregated_data.html

For a breakdown of how the EM algorithm works "under the hood", please see the vignette available in this repository:

• em_explanation.Rmd