JMIR Publications

ABSTRACT

Background:

The Global Burden of Disease (GBD) study provides the most comprehensive cross-national burden estimates available, but translating GBD data downloads into publication-ready analyses requires an integrated, multi-method workflow spanning trend estimation, changepoint detection, demographic decomposition, forecasting, frontier benchmarking, and health inequality measurement. Existing tools cover portions of this workflow but none integrates it end to end in a code-free interface, creating a disproportionate barrier for investigators in resource-limited settings.

Objective:

To develop and release an open-source R Shiny dashboard, the GBD Analysis Suite, that integrates the 11 analytical modules required for a standard GBD burden of disease study into a single user-uploadable, code-free web interface, and to describe its architecture, methodology, and intended use.

Methods:

The application was built using R 4.5.3 and the Shiny framework, integrating established statistical packages (forecast, segmented, openxlsx, ggplot2, plotly, and others) into 11 modules covering data import, choropleth mapping, estimated annual percentage change (EAPC), joinpoint regression, Kitagawa and Das Gupta decomposition, multi-model time-series forecasting, descriptive visualization, Socio-demographic Index (SDI) based frontier analysis, age period cohort (APC) visualization, slope and relative index of inequality (SII and RII), and batch export. The application accepts any GHDx-formatted CSV or Excel file, auto-detects standard column names, and enforces internal consistency by construction. Kitagawa decomposition uses the symmetric mid-point formulation to guarantee exact additivity, and all forecast models use Box-Cox log transformation with bias-corrected back-transformation to guarantee non-negative point estimates and prediction intervals.

Results:

The application is deployed at https://bela2003.shinyapps.io/GBD-Analysis-Suite/ and runs in any modern web browser without local installation. On the bundled 204-country, 34-year demonstration dataset, a complete batch export (an 8-sheet Excel workbook plus 10 or more figures) completes in approximately 90 seconds at a peak memory footprint of approximately 800 MB. A feature comparison against five existing tools (GBD Compare, GBD Foresight, NCI Joinpoint, BAPC, NordPred) across 14 capability dimensions shows that the GBD Analysis Suite is the only reviewed platform combining user-uploaded data input with code-free access to all six core analytical methods of modern GBD papers. Source code is publicly available at https://github.com/levai-2003/GBD-Analysis-Suite under the MIT license.

Conclusions:

The GBD Analysis Suite lowers the technical barrier to reproducible, multi-method burden of disease analyses. It is designed particularly for researchers in low- and middle-income countries where biostatistical support, commercial software licenses, and high-performance computing are often unavailable. The modular design imposes no structural limitations on disease area, geography, or time window.