Global UPF Supply Atlas and PIF Burden Estimation Dataset (2010–2023)

This dataset provides the first open, cross-validated global atlas of ultra-processed food (UPF) dietary energy supply shares for 174 countries (2010–2023), together with time-varying potential impact fraction (PIF)-based burden estimates for type 2 diabetes and cardiovascular disease (2010–2021). A five-layer sensitivity cascade transparently decomposes compound uncertainty across exposure imputation, supply-to-intake mapping, dose-response transportability, food waste deduction, and regional atlas validity. Core finding: PIF-based potentially avoidable burden was 9.53 million DALYs in 2021 (credibility-weighted: 3.62M; range: 1.59–9.53M). Burden growth from 2010 to 2021 (+33%) was 103% denominator-driven (population ageing, NCD prevalence expansion) and −3% exposure-driven, consistent with a 20-year generational lag between dietary-structure change and population obesity.