Deep learning four decades of human migration

A global annual migration-flow dataset spanning 1990 to 2024 has been produced using deep-learning models, as published in Nature on June 10, 2026. This dataset estimates human movements across 230 countries, offering improved temporal resolution, broader coverage, and more precise uncertainty estimates compared to previous methodologies. The research leverages diverse data sources, integrating them through advanced machine learning techniques to reconstruct migration patterns over 34 years. This comprehensive dataset is expected to significantly enhance the understanding of global demographic shifts and their underlying drivers. The study's findings provide a granular view of how populations have moved internationally, offering valuable insights for policymakers, researchers, and international organizations.