analysis2.Rmd

---
title: "Analysis"
csl: the-american-naturalist.csl
output:
  html_document:
    theme: cerulean
    toc: yes
  pdf_document:
    toc: yes
<!-- bibliography: references.bib -->
editor_options: 
  chunk_output_type: console
---

<!--
IMAGES:
Insert them with: ![alt text](image.png)
You can also resize them if needed: convert image.png -resize 50% image.png
If you want to center the image, go through HTML code:
<div style="text-align:center"><img src ="image.png"/></div>

REFERENCES:
For references: Put all the bibTeX references in the file "references.bib"
in the current folder and cite the references as @key or [@key] in the text.
Uncomment the bibliography field in the above header and put a "References"
title wherever you want to display the reference list.
-->

<style type="text/css">
.main-container {
  max-width: 1370px;
  margin-left: auto;
  margin-right: auto;
}
</style>

```{r general options, include = FALSE}
knitr::knit_hooks$set(
  margin = function(before, options, envir) {
    if (before) par(mgp = c(1.5, .5, 0), bty = "n", plt = c(.105, .97, .13, .97))
    else NULL
  },
  prompt = function(before, options, envir) {
    options(prompt = if (options$engine %in% c("sh", "bash")) "$ " else "> ")
  })

knitr::opts_chunk$set(margin = TRUE, prompt = TRUE, comment = "", message = FALSE,
                      collapse = TRUE, cache = FALSE, autodep = TRUE, warning = FALSE,
                      dev.args = list(pointsize = 11), fig.height = 3.5,
                      fig.width = 4.24725, fig.retina = 2, fig.align = "center")

options(width = 137)
```

<!--
model 1: risk factor for infections univariate
Size of farm
age
helminth
experience of farmer
ratio

model 2: prediction
prediction of the etiology based on symptoms
resp, diarrhea, cns, anemia, foot lesions : 5 symptoms
also 3 variables: time since onset, mortality, morbidity

Hi Marc,
As we discussed, I would like to send you data for Modelling.
It includes 2 sheets with variables of 2 models:
1.       Risk factor Analysis(6 variables):
a.       Age
b.      Size of flock
c.       Other chicken flock
d.      Farmer experience (years)
e.      Ratio weight/ normal weight
f.        Helminth (Yes/No)
2.       Patterns to predict disease (8 variables):
a.       Respiratory
b.      Diarrhea
c.       Anemia
d.      CNS
e.      Joint/Foot problems
f.        Mortality
g.       Morbidity
h.      Days since onset
Let me know if anything is not too clear. Thanks a lot for your support!
Best regards,
Vân
-->

## Packages

```{r}
library(dplyr)
```

## Risk factor analysis

Reading the first tab of the excel file and putting in shape:

```{r}
rfa <- readxl::read_excel("DataforMarc.xlsx", "Risk Factors analysis") %>% 
  mutate_at(vars(OthChickFlock, `Helminth Y/N`, ends_with("FINAL")), ~ . > 0) %>% 
  select(-Farmcode)
```

Let's perform nivariable analyses with logistic regressions. It's important to
note that these models are only for prediction, in the specific context of the
chicken farms of this study, not for explanation. The following function
returns the effect of variable`x` on variable `y`

```{r}
x_effect <- function(x, y) {
  paste0("`", y, "`~`", x, "`") %>% 
    formula() %>% 
    glm(binomial, rfa) %>% 
    summary() %>% 
    coef() %>% 
    data.frame() %>% 
    `[`(-1, ) %>% 
    data.frame("risk_factor" = rownames(.), ., stringsAsFactors = FALSE)
}
```

Let's use this function for all `x` and `y` variables:

```{r}
for (y in grep("FINAL", names(rfa), value = TRUE)) {
  print(y)
  print(bind_rows(lapply(grep("FINAL", names(rfa), value = TRUE, invert = TRUE), x_effect, y)))
}
```

Interpretation:

* the risks of AVI, ORT and FLU infection increase with age;
* the presence of helminths decreases the risk of GUM;

Note however that if we had to correct for multiple tests (54), a
significativity threshold of 0.05 would become 0.00093 and none of these tests
would be significant.

```{r include = F, eval = F}
efs <- readxl::read_excel("DataforMarc.xlsx", "Patterns to predict disease") %>% 
  select(-Farmcode) %>% 
  mutate_at(vars(-matches("^Cu|^Pr")), ~ . > 0)
```