First output for Blueprint

[athowes]

Document that describes:

1. data sources
2. model inputs
3. model structure
4. basic model outputs (e.g. some epidemic curves)
5. first experiments (e.g. heatmaps showing how varying farUVC characteristics influences epidemic outcomes).

Around 2-4 pages total including figures. First half can be built on top of the vignette https://mrc-ide.github.io/helios/articles/model.html.

For the second half, some next steps are:

• [ ] Determine parameter sweeps to run. Which parameters, which range of values, whether univariate or multivariate
• [ ] Determine outcome measures used (total infections, total deaths)
• [ ] Write code to do parameter sweeps with a suitable number of replications
  • Power analysis for this?
  • Use hypercow to run analysis. See #41.
• [ ] Save analysis results (maybe big files could be a problem?)
• [ ] Write code to generate graphical summaries of parameter experiments
• [ ] Write code to generate tabular summaries of parameter sweep experiments

[cwhittaker1000]

Format - .Rmd :)

@athowes - Flesh out first half of the doc (model description, input data description and graphical visualisation, model parameters and inputs description) - with a particular focus on the graphs, with the text being second priority after those are complete.

@tbreweric - Generating the exemplar model outputs and the parameter sweeps. For the exemplar model outputs, pick an R0 e.g. R0 = 2, with a beta split such that you have 30% transmission in the household, 20% in schools, 20% in workplaces, 20% in leisure settings, and 10% in community. And then run three simulations - one of which is without farUVC, and the other of which is with farUVC, installed at a coverage of let's say 50%, and do this for both random and targeted, across all structures except households to begin with. Do this for the epidemic scenario setting to begin with. For the parameter sweeps, do R0 = {1.25, 1.5, 2, 2.5, 3} and farUVC coverage do it between 0 and 1, farUVC efficacy between 0 and 1, both in increments of 0.2; and do this for both targeted and random coverage types. In all cases, run 5 replicates and calculate the total number of individuals infected. Then calculate infections averted as total_without_farUVC - total_with_farUVC.

[athowes]

Suggest putting Adam work in helios/vignettes/blueprint1.Rmd and Tom work in helios/vignettes/blueprint2.Rmd then merging into helios/vignettes/blueprint.Rmd once both finished (to avoid any merge issues while working on them).

[athowes]

Suggest script to determine $R_0$ based on parameter settings to be saved somewhere for future use. (Perhaps could be a function?)

[athowes]

• 5 $R_0$ times 7 coverage times 7 efficacy times 2 settings = 350 runs
• Each run takes (10,000 population, 150 days, timestep of 0.5) 1-2 minutes
• So total around 6 - 12 hours (for one replicate)

[athowes]

Shipped! Closing this.