This repository hosts the code, data and supporting analysis for “Time is of the essence: effectiveness of dairy farm control for H5N1 is limited by fast spread.” The analysis was structured in the form of an R package to facilitate reproducibility.

You can install the development version of rRsurveillance from GitHub with:

# install.packages("remotes")
remotes::install_github("wf-id/h5n1speed")

The primary analysis scripts are available as:

• R/generate_outbreak_chart.R contains the code used to generate Figure 1, a figure of the ongoing H5N1 outbreak (available in manuscript/figures/outbreak.pdf)
• manuscript/delay_strategy.qmd contains the code to generate Figure 2, panel A and assembles the figures to make Figure 2.
• manuscript/intervention-effectiveness.R contains the code to generate the figures shown in Figure 3 panels A-C which shows the effectiveness of each surveillance strategy and detection threshold at prevent infections.
• manuscript/time-to-effective-strategy.R contains the code to generate the time required to realize an effective strategy. These values are used to generate Figure 3 panels D-F.
• dev/test-asymptomatic.R contains the code to explore the role of asymptomatic transmission on detection and prevention of an outbreak and generates data for Figure 4A
• dev/asymptomatic-test.R contains the code to explore the contribution of asymptomatic transmission and delay in terms of relative risk and also generated the complted Figure 4 with panels B-C.
• dev/conceptual.R contains the code to generate the outputs used to make the conceptual figures shown in Figure 5.

Supplmental analysis:

• dev/test-mortality.R contains the code to explore the role of disease induced mortality on detection and prevention of an outbreak.

Please note that the R scripts were run across 38 threads on a single node of the Wake Forest University HPC which took ~5-6hrs. Running these scripts on a single machine may take a significantly long time.

We have parameterized our analysis into an R package to faciliate ease of use and allow us to explore scenarios of speed on surveillance and intervention strategies.

• run_det_ode contains the code used to run a deterministic compartmental model as described in our manuscript. This code use the deSolve package to solve the differential equations. Additionally, this code adds in the milk production.
• detect_infections_ode processes the outputs from the run_det_ode and find when some particular detection threshold has been met (e.g., number of cows infected cumulatively, number of symptomatic cows with clinical signs, some proportion drop in milk production). This function returns a list object with the timepoint at which the detection threshold is met.
• run_intervention_ode allows a given intervention to be applied given some detection and surveillance approach (e.g., the number symptomatic detected) at that time with some delay with varying intensity.

As a reminder, the ODE is shown below:

$$ \begin{align} \frac{dS}{dt} &= - \beta S I/N \\ \frac{dI}{dt} &= \beta S I/N - \gamma I \\ \frac{dB}{dt} &= \gamma I - \kappa B \\ \frac{dR}{dt} &= \kappa B \\ \end{align} $$

Milk production is calculated as follows:

$$ \text{Milk} = 100 * (S + I) + 75 * B + 80 * R $$