Recent Releases of CREAHIA
CREAHIA - 0.7.4
Summary
We moved the exposure map logic to creaexposure instead, so that we can more easily rely on new maps.
- Updated
get_conc_baseline_*()functions to usecreaexposure::get_concentration() - Added
.validate_unit()— errors if the returned raster has no unit or unexpected unit - Removed dead
no2_min_incrparameter (was a no-op since oldget_conc_baselinewas replaced) - Added
creaexposureas a dependency in DESCRIPTION
Breaking changes
get_conc_baseline()andget_conc_baseline_no2()no longer acceptno2_min_incr
Natural Resources - Air Quality
- R
Published by hubert-thieriot 3 months ago
CREAHIA - 0.7.2
GBD2023 Epidemiological data
MAJOR: This version includes GBD 2023 epidemiological data.
Other improvements include:
- improved testing (e.g. immutability tests cover more parameter sets)
- ihme renamed to epi_count_long (and epi to epi_rate_wide) for clarity
Natural Resources - Air Quality
- R
Published by hubert-thieriot 7 months ago
CREAHIA - 0.7.1
Major Refactoring
Nothing almost changed in the computation. Mostly code structure and naming changes.
- renaming epidemiological variables/files e.g. epi_rate_wide and epi_count_long vs epi_for_gbd and ihme
- lower case cause, outcome
- age_weights clarification and testing
- caching ihme (i.e. epi_count_long)
- cleaned jonction/split cause_outcome
- cleaned some epi data generation
- improved documentation
Natural Resources - Air Quality
- R
Published by hubert-thieriot 7 months ago
CREAHIA - 0.7.0
0.7.0
MAJOR CHANGE with breaking modifications
Streamlined paf x epi computation
The separation between PAF computation and Impacts computation is now effective. In previous versions, some paf computation still happened in the epi part.
Moreover, while PAF for RR curves and CRFs are computed separately, they are applied together to epidemiological data (whilst they were applied in different places in previous versions).
Column naming
Most columns are now lower case + underscores.
CRFs definition
Causes, pollutant and outcomes better defined in crfs files (rather than relying on underscores)
Improved testing
test_results_immutability now part of CI, meaning new version results will always be compared to old ones. It now also has a better system for "allowed breaking results".
Natural Resources - Air Quality
- R
Published by hubert-thieriot 8 months ago
CREAHIA - 0.6.0
Key changes
Viscusi updated
Visusci valuation now uses GNI Atlas by default
New valuation methods
- User can now access valuation in different countries/years using
get_valuations(iso3s, years)method. - No more intermediate stage with 2017 International Dollars
- Valuation methods works for countries that don't have GNI per capita in constant USD (only if elasticity = 1)
Income vs GDP
get_income() replaces get_gdp() as it also includes GNI.
Others
get_econ_forecast() arguments renamed: forecast_years, reference_year, use_gdp_scaling, discount_rate. Its Scaling logic clarified: fatal outcomes (Deaths, YLLs) use death-rate scaling; non-fatal use population scaling.
Improved tests and maintenance
Natural Resources - Air Quality
- R
Published by hubert-thieriot 9 months ago
CREAHIA - Release 0.4.3
0.4.3
This version radically changes (i.e. improves) the way confidence intervals are being computed in paf and then in paf * epi.
See doc/paf_epi_uncertainty and doc/paf_epi_uncertainty for more details.
Central values have not been affected by this update.
Adjusting confidence intervals crossing zero
To ensure consistency in our analysis, we adjust the confidence intervals of the Population Attributable Fraction (PAF) and the associated effects when they span both negative and positive values.
Specifically, if the central PAF estimate is positive, we set the lower bound of the confidence interval to zero, effectively removing any negative values. Conversely, if the central estimate is negative, we set the upper bound to zero, excluding positive values. This approach ensures that the confidence intervals align with the direction of the exposure effect, preventing the inclusion of implausible PAF values.
Natural Resources - Air Quality
- R
Published by hubert-thieriot over 1 year ago
CREAHIA -
New confidence interval computaiton
For GBD2019
-
Before:
- Confidence interval (CI) of OthResp and OthCV derived by substracting the CI of IHD/Stroke/COPD from CI of TotalResp of TotalCV, for all age groups above 25
-
After:
- Confidence interval (CI) of OthResp and OthCV derived by summing the variance of all other respiratory/cardiovascular diseases from GBD, for the single GBD age group 25+
For GBD2017
- Before:
- Same as GBD2019
- After:
- Because we can’t seem to access GBD2017 results anymore, I can’t download all other resp/cv diseases data. Instead, I’ve just updated the old method (e.g. upper = upper(total)-sum(IHD|STROKE|COPD)) to the new one (variance(Total) = variance(IHD|STROKE|COPD) + variance(Others))
Natural Resources - Air Quality
- R
Published by hubert-thieriot over 1 year ago