Liu J, Cai Z, Gustafson P, McDonald DJ (2024) rtestim: Time-varying reproduction number estimation with trend filtering. PLOS Computational Biology 20(8): e1012324. https://doi.org/10.1371/journal.pcbi.1012324
This is a project of application on time-varying reproduction number estimation using convex optimization and proximal algorithm solvers.
We built a nonparametric regression model with trend filtering penalty to estimate instantaneous reproduction number, a key tool to reveal the transmissibility of infectious diseases.
We developed a method using proximal Newton method, ADMM, and dynamic programming to solve the problem.
We wrote the main manuscript and all supplementary documents and conducted the experimental study.
We developed a lightweight and computationally efficient R package rtestim (coded from scratch).
Source Code
- All related data can be found in this GitHub repository.
Software Development
- R package for time-varying reproduction number estimation for infectious disease rtestim with core algorithms developed in C++.
Toolkit for the Project
- C, C++, R languages.
- R Packages: Rcpp, RcppArmadillo, RcppEigen, cpp11, ggplot2, batchtools, etc.
- Linux.
- Computing Resource: Compute Canada.
- R Studio, VScode.
- LaTeX, RMarkdown.