2026-10-10T16:38:35Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33717313 2026-09-14T11:31:43Z category_27193 category_27190 item_type_3 month_year_09_2026

Analysis code for "The 'Development Paradox' in Global Diabetic Retinopathy Burden: A Panel Data Study on Health Expenditure Structure, AI Readiness, and Ophthalmologist Density Thresholds (1990–2021)" Menglan Zhu (23953533) Health promotion Health equity diabetic retinopathy global burden of disease health expenditure ophthalmologist density <p dir="ltr">This item contains the complete analysis code package (Supplementary File 2) accompanying the manuscript: The "Development Paradox" in Global Diabetic Retinopathy Burden: A Panel Data Study on Health Expenditure Structure, AI Readiness, and Ophthalmologist Density Thresholds (1990–2021).</p><p><br></p><p dir="ltr">The study examines the global burden of diabetic retinopathy (DR) across 204 countries and territories (1990–2021) and its national-level associations with health expenditure structure, ophthalmologist workforce density, and artificial intelligence (AI) readiness, using panel data methods with fixed effects, instrumental variable estimation, and Hansen threshold regression.</p><p><br></p><p dir="ltr">The code package consists of nine scripts (R and Python) together with a README describing the runtime environment and execution order:</p><p dir="ltr">- 01_data_extraction.R — extract DR outcomes from GBD 2021 raw files</p><p dir="ltr">- 02_data_merge_imputation.R — construct the analysis panel; linear interpolation and SDI-quintile median filling</p><p dir="ltr">- 03_panel_models.R — fixed-effects, lagged, and instrumental-variable panel models (Table 2)</p><p dir="ltr">- 04_threshold_crosssectional.R — Hansen threshold regression, interaction models, and generalized additive models (Figure 3)</p><p dir="ltr">- 05_inequality_analysis_legacy.R — legacy region-level inequality analysis (superseded, retained for provenance)</p><p dir="ltr">- 06_sensitivity_diabetes_prevalence.R — fixed-effects sensitivity analysis adjusting for national diabetes prevalence</p><p dir="ltr">- 07_inequality_sii_recompute.py — country-level slope index of inequality and concentration index (n = 204; 1,000 bootstrap replications)</p><p dir="ltr">- 08_figures_2B_4.py — regenerate Figures 2B and 4 (600 dpi)</p><p dir="ltr">- 09_relative_change_ui.py — 95% uncertainty interval of the 1990–2021 relative change from GBD draws</p><p><br></p><p dir="ltr">Runtime environment: R 4.5.x (dplyr, sandwich, lmtest, boot, mgcv, readr) and Python 3.12 (pandas, numpy, scipy, matplotlib).</p><p><br></p><p dir="ltr">The underlying data are publicly available from the Global Burden of Disease Study 2021 (IHME GBD Results Tool), the WHO Global Health Expenditure Database, the International Council of Ophthalmology Atlas, and the Oxford Insights Government AI Readiness Index (2024 edition). Raw data are not redistributed with this package per the providers' terms of use; download paths and extraction parameters are documented in the script headers and the README.</p> 2026-09-14T11:31:43Z Dataset Dataset 10.6084/m9.figshare.33717313.v1 https://figshare.com/articles/dataset/Analysis_code_for_The_Development_Paradox_in_Global_Diabetic_Retinopathy_Burden_A_Panel_Data_Study_on_Health_Expenditure_Structure_AI_Readiness_and_Ophthalmologist_Density_Thresholds_1990_2021_/33717313 CC BY 4.0