The package cityClimateHealth makes it simple to estimate heat-health impacts at small spatial scales. Starting from a messy exposure and outcome dataset, we can quickly estimate heat-health impacts.
Installing the package
remotes::install_github("cmilando/cityClimateHealth")you may have to first install STAN and cmdstanr:
install.packages('cmdstanr', repos = c('https://stan-dev.r-universe.dev', getOption('repos')))Usage
This package can be used in three main ways:
| 1-stage design | 2-stage design | Spatial Bayes |
|---|---|---|
A 1-stage conditional poisson model when estimating a single set of beta coefficients for heat-health impacts across single or multiple zones: vignette("one_stage_demo")
|
A 2-stage design is used when estimating heat-health impacts across many zones, but where individual zone models are desired: vignette("two_stage_demo")
|
If numbers are very small in the 2-stage design, spatial bayesian methods can be used to tighten confidence intervals: vignette("bayesian_demo")
|
In implementations, an attributable number calculation is applied to model outputs, see vignette("attributable_number").
Time-series functions are in-progress
Starting a new analysis
To start a new analysis, you will need the following 4 datasets:
| Exposure | Outcomes | Populations | Spatial |
|---|---|---|---|
Exposures at the daily scale for each geo_unit
|
Health outcomes at the daily scale for each geo_unit
|
Population data for each subdivision of the health outcome data that you want results for | A map showing how the various geo_units are neighbors |
This package comes pre-loaded with sample datasets of each type (e.g., ma_exposure, ma_deaths, ma_pop_data, and ma_towns respectively) so each of these methods can be explored.