EpiInvert

REF	Alvarez et al. (2021)
Docs	lalvarezmat.github.io/EpiInvert/
Github	github.com/lalvarezmat/EpiInvert
Last commit	Dec, 2023
Installation	via devtools

Brief description

Brief summary of the method from the paper

EpiInvert is an epidemiological method that estimates the time-varying reproductive number and restores incidence curves by inverting the renewal equation using variational techniques. The approach corrects biases introduced by reporting inconsistencies, including weekly and festive biases, ensuring robust epidemic trend estimation. EpiInvert estimates R(t) by inverting the renewal equation using signal processing techniques, providing a reliable measure of epidemic dynamics. It corrects systematic underreporting due to weekends and holidays by detecting anomalies based on historical trends, redistributing cases across affected days to reduce artificial fluctuations, and adjusting Rt estimates to reflect true transmission patterns. It also includes a forecasting model that predicts epidemic trends using historical trends.

Methods

This package modifies fixed sliding windows, using a variational method that includes solving R(t) using the classic Wallinga Teunis method (Wallinga and Teunis (2004)).

Assessment

Features
Ability to nowcast/forecast	Yes, Use ‘EpiInvertForecast’ for forecasting
Incorporates delay distributions	No
Estimates expected cases	Yes
Communicates uncertainty	Yes
Validation
Documentation of package methods	Yes
Documentation of package implementation	Yes

Sample code

See this vignette for an example of forecasting, and this vignette for a comparison between EpiInvert and other related packages.

Alvarez L, Colom M, Morel J-D, Morel J-M. 2021. Computing the daily reproduction number of COVID-19 by inverting the renewal equation using a variational technique. Proceedings of the National Academy of Sciences 118:e2105112118; doi:10.1073/pnas.2105112118.

Wallinga J, Teunis P. 2004. Different epidemic curves for severe acute respiratory syndrome reveal similar impacts of control measures. American Journal of epidemiology 160: 509–516.