In choisy/viparc: Utilitary functions for the ViParc project

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
})

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

Packages

library(viparc)
library(dplyr)

Heatmap figure

Heatmap of the surveillance disease and drug usage in several farms for a number of weeks:

data("chcknfarms")
chcknfarms %>%
  filter(CYCLE == 1) %>%
  transmute(space   = FARMCODE,
            time    = WEEK,
            disease = SRespiratory,
            drug    = AntibioticUse) %>%
  heat_map()

Note that the input data frame to heat_map should have 4 variables named space, time, disease and drug.

Identifying weeks with no AMU

Let's consider for example the following vector of AMU:

amu <- rep(rep(c(TRUE, FALSE), 6), c(7, 2, 1, 2, 3, 1, 2, 1, 1, 1, 1, 3))

See this example:

data.frame(amu) %>%
  mutate(
    with_amu = amu,
    wout_amu = no_amu(amu, 1)) %>%
  select(-amu)

Note that, as expected, the only combination that we cannot have on any row is c(TRUE, TRUE). Then, if one wants to discriminate any week either into AMU or no-AMU, the code will have to be:

data.frame(amu) %>%
  mutate(
    with_amu = amu,
    wout_amu = no_amu(amu, 1)) %>%
  select(-amu) %>% 
  filter(with_amu + wout_amu > 0) %>% 
  transmute(amu = with_amu)

Looking for the effect of AMU on disease

Let's consider a data set that could look like this:

(dataset <- data.frame(
  amu = c(rep(TRUE, 4), rep(FALSE, 5), TRUE, rep(FALSE, 6), TRUE,
          FALSE, TRUE, rep(FALSE, 4), TRUE, FALSE),
  disease = c(rep(TRUE, 3), FALSE, TRUE, FALSE, rep(TRUE, 2), FALSE,
              rep(TRUE, 2), rep(FALSE, 3), rep(TRUE, 2), rep(FALSE, 3),
              TRUE, rep(FALSE, 2), rep(TRUE, 3))
))

Among the weeks with no diseases, we want to compare 2 groups of weeks:

the weeks $i$ with no AMU as well as no AMU either on week $i - 1$;
the weeks with AMU

and for each of these 2 groups we want to count the number of weeks $j$ for which a disease was observed at week $j + 1$ or $j + 2$.

To make it clear, note that here we are thus eliminating from the analysis

all the weeks $k$ with no AMU but with AMU on week $k - 1$ and
all the weeks with disease.

The R code for that would be:

dataset %>% 
  mutate(no_amu = no_amu(amu, 1),              # no AMU on weeks i-1 and i
         diseas = any_disease(disease, 2)) %>% # disease on weeks i+1 or i+2
  na.exclude() %>% 
  filter(! disease,                            # removes weeks with disease
         amu + no_amu ) %>%                    # removes weeks with no AMU but...
  mutate(amu = ! no_amu) %>%                   # ... with AMU the week before.
  select(-disease, -no_amu)  %>% 
  table() %>% 
  fisher.test()

These piece of code in basically included in the amu_causes_disease() function. See the examples below:

amu_with_diseases %>%
 amu_causes_disease()

amu_with_diseases %>%
 amu_causes_disease(disease = "DIARRHOEA")

amu_with_diseases %>%
 mutate(Sick_yes = DIARRHOEA | RESPIRATORY ) %>%
 amu_causes_disease()

amu_with_diseases %>%
 mutate(Sick_yes = DIARRHOEA | RESPIRATORY ) %>%
 amu_causes_disease(week_disease = 3)

Indexing the week in a sequence

See how functions time_index() and fill_gaps() can be used to do the job:

dataset %>%
  mutate(amu_week = time_index(amu),
         disease2 = fill_gaps(disease),
         disease2_week = time_index(disease2)) %>% 
  select(amu, amu_week, disease, disease2, disease2_week)

Then, you can extend this in a longer pipeline:

dataset %>%
  mutate(amu_week = time_index(amu),
         disease = fill_gaps(disease),
         disease_week = ifelse(disease, time_index(disease), NA),
         disease_start = disease_week < 2,
         amu_event = ifelse(amu,
                            ifelse(amu_week > 1, "D", "S"),
                            ifelse(amu_week > 1, "N", "A")))