---
title: "Getting started"
output: html_document
vignette: >
  %\VignetteIndexEntry{Getting started}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

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

```{r setup}

library(tidyverse)
library(testerror)

options(mc.cores = 2) # parallel::detectCores())
rstan::rstan_options(auto_write = FALSE)

here::i_am("vignettes/testerror.Rmd")
source(here::here("vignettes/formatting.R"))

```

# Correcting panel tests for test error

We define panel testing as testing for multiple subtypes of disease, where any
positive result for a subtype is regarded as a positive for the disease. As we
describe in our paper this leads to compound error, which is particularly
significant when the individual components are present at low prevalence and the
number of components involved is large.

In the following example a simulation a population with true prevalence of 15%,
is tested for using a panel of 15 tests each with sensitivity of 0.99 and
specificity of 0.8. The individual components in the panel each generate small
amounts of error which compound to test positivity (red) being a serious
overestimate of panel prevalence (blue).

This package provides modelled estimates of prevalence from test and panel test
positivity.

```{r}
tmp = testerror:::panel_example(
  panel_prev = 0.15, comp_spec = 0.99, 
  comp_sens = 0.8, n_comp = 15, n_samples = 2000, 
  n_controls = 800, n_diseased = 30, exact_controls = TRUE) #, exact_samples = TRUE)

corr2 = true_panel_prevalence(
   test_results = tmp$samples %>% dplyr::select(id,test,result = observed),
   false_pos_controls = tmp$performance$false_pos_controls,
   n_controls = tmp$performance$n_controls,
   false_neg_diseased = tmp$performance$false_neg_diseased,
   n_diseased = tmp$performance$n_diseased,
   method = "lang-reiczigel"
   #method = "rogan-gladen"
   #method = "bayes"
)

testerror:::demo_bar_plot(tmp$summary %>% dplyr::inner_join(bind_rows(corr2),by="test"),.pcv=FALSE)
```

These functions are elaborated in the other vignettes.