---
title: "Published Test-then-Vaccinate ROI Analyses"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{published-analysis}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

We computed the analysis in [this publication](https://doi.org/10.1098/rsif.2019.0234) in a series of steps, each dependent on the next, linked together via the traditional development tool [`gnu make`](https://www.gnu.org/software/make/manual/html_node/index.html).  We fit some example serological data sets (Morrison 2010 and L'Azou 2016, both collected from published studies and available as part of this package),
and then use those fits to simulate populations.

All of the scripts mentioned in this vignette are included in the `inst/extdata` of this package, and can be located wherever your R packages are installed, or accessed via the `build.project` tool to make a skeleton, or by using `list.files(system.file("extdata/pub", package="denvax"), pattern=".R", full.names = T)` to get the paths.

The skeleton includes a 'Makefile' that defines a general set of dependencies, as well as the specific steps used to generate the figures in the published manuscript.  The `simple.R` script demonstrates a simple start-to-finish analysis (using the Morrison 2010 data included in the package).

In either approach, the steps are:

1. develop fit(s) for local serological data (e.g., use the `fit.R` script from `build.project` in Rstudio, write your own script with the `denvax::serofit` function, or use make with the target `%-fit.json` in Makefile).
2. use a fit to synthesize a population (e.g., `synthesize.R` script, the `denvax::synthetic.pop` function, or the target `%-lh.rds` in Makefile)
3. use the synthetic population to estimate the lifetime outcome probabilities (e.g., `digest.R` script, the `denvax::npxa` function, or the target `%-npxa.rds` in Makefile)
4. combine the probabilities with the desired boundaries for interventions (initial testing age and maximum number of tests) and to compute the ROI equation coefficients (`denvax::ROIcoeffs` function)
5. finally, combine the coefficients with desired cost fraction ranges (i.e., the cost of vaccination and testing as fractions of the estimated burden of secondary infection, or $\nu$ and $\tau$ respectively), to determine the threshold region for positive ROI (`denvax::ROI` function).

Within the `pub/` directory, we have another makefile and scripts which cover the specific analyses used in the publication.