Google Research has unveiled the Population Dynamics Foundation Model (PDFM), a geospatial AI system designed to address critical data gaps in global public health. By synthesizing privacy-preserving signals—including search trends, human mobility patterns, and environmental data—into monthly-refreshed location embeddings, PDFM enables epidemiologists to enhance disease surveillance and outbreak response without building custom models from scratch. The research team, in partnership with five global health institutions, demonstrated PDFM's effectiveness across diverse public health challenges. Results included a 36% improvement in explaining variance for MMR vaccination cross-border patterns, mortality nowcasting for cardiovascular disease matching census-based models, and statistically significant advances in dengue outbreak forecasting across Mexican municipalities. The model also showed promise in predicting postpartum depression risk and forecasting cholera emergence in resource-constrained settings like the Democratic Republic of the Congo. Unlike traditional epidemiological approaches that require extensive task-specific data collection, PDFM's location embeddings function as plug-and-play inputs to existing health science workflows. This paradigm shift could prove especially valuable during rapid disease outbreaks and in regions with limited surveillance infrastructure, where timely geospatial context has historically been difficult to obtain.