A Machine-Learning-Imputed Global Atlas of Ultra-Processed Food Supply Shares and PIF-Based Burden Estimates for Non-Communicable Diseases

Researchers created the first global atlas of ultra-processed food supply shares for 174 countries from 2010 to 2023 using machine learning algorithms. The atlas maps 44 published national estimates based on food supply structure. The study estimates that ultra-processed foods caused approximately 3.6 million disability-adjusted life years in 2021. Between 2010 and 2021, the increase in health burden was primarily driven by population ageing and disease prevalence expansion, while changes in food supply contributed negatively by 3 percent. The research suggests that the ultra-processed food environment was already established in most countries by 2010. All data, methodology, and code were released as public health infrastructure.