Geographical heterogeneity in interpretive and screening robustness of PM₂.₅ mortality decomposition across 204 populations

Description

This dataset provides the derived data, diagnostic outputs, figures, and reproducibility materials supporting the study “Geographical heterogeneity in interpretive and screening robustness of PM₂.₅ mortality decomposition across 204 populations.” The study decomposes the change in ambient PM₂.₅-attributable mortality between 2000 and 2023 across 204 countries and territories into population growth, population ageing, and the composite attributable mortality rate. Contributions are evaluated over all six replacement orders. The package reports symmetric order-robust contributions and complementary diagnostics of interpretive sensitivity, including sign changes, changes in the largest absolute contributor, and order-induced uncertainty widths. This version adds the analyses developed during manuscript revision: alternative UN M49 geographic classification; population-threshold sensitivity analyses; 10-group and 5-group age-resolution checks; task-specific screening comparisons based on mean attributable burden, final attributable burden, raw attributable deaths, and the direction of change; and exploratory spatial-dependence diagnostics. The spatial extension uses frozen Natural Earth 1:50m geographic inputs, k-nearest-neighbour spatial weights (k = 4, 8, and 12), Moran’s I statistics, unrestricted and within-region permutation tests, and Holm adjustment across the full family of 16 tests. The deposited materials consist of: (1) a machine-checkable reproducibility package containing scripts, frozen derived inputs, geographic resources, expected outputs, checksums, and verification routines; (2) a supplementary data workbook containing the complete tabular results; and (3) supplementary information documenting the methods, diagnostic definitions, sensitivity analyses, and supporting figures. The package reproduces the analyses from supplied derived data without requiring access to the restricted source archives. Optional source-level reconstruction requires the five licensed GBD 2023 archive files listed in the README. Those source archives are not redistributed. All reported results are based on point estimates and should not be interpreted as GBD uncertainty intervals, causal effects, or direct measures of intervention need.

Steps to reproduce

1. Download ASAP_Reproducibility_Package.zip and extract it to a local directory. 2. Open a terminal in the extracted package directory. Python 3.10 or later is recommended. 3. Install the required Python packages: python -m pip install -r requirements.txt 4. Reproduce the primary decomposition and robustness checks from the supplied derived data: python run_review_checks.py --outdir review_outputs 5. Reproduce the geographic, population-threshold, age-resolution, screening-task, and spatial-dependence extensions: python run_spatial_extension.py --outdir extension_outputs 6. Regenerate the manuscript and supplementary figures: python make_figures.py --outdir figures 7. Run the complete package-integrity and numerical verification suite: python verify_package.py The verification routine checks file hashes, primary decomposition results, regenerated extension tables, age-aggregation closure, spatial-neighbour structures, Moran’s I calculations, permutation results, and Holm-adjusted p-values. A small built-in boundary-case test can also be run independently: python run_reproducibility.py --self-test Optional source-level reconstruction requires the five licensed GBD 2023 source archives identified by filename and checksum in the README. Place those archives in a local directory and run: python run_reproducibility.py --input-dir /path/to/gbd_archives --output-dir source_outputs --package-dir . python run_spatial_extension.py --input-dir /path/to/gbd_archives --outdir extension_source_outputs The original GBD archives are not included because their redistribution is restricted. They are not required for the standard derived-data reproduction route.

Institutions

• Kocaeli Üniversitesi

  Kocaeli, İzmit

Categories

Licence