Unlocking Earth AI’s planetary geospatial foundation models for global public health
Google Research
Google Earth AI introduced Population Dynamics Foundation Model (PDFM) embeddings as plug-and-play geospatial inputs for public-health ML workflows across multiple diseases and geographies. In evaluations, PDFM improved MMR vaccination coverage explained variance for 146 U.S.-Canada border counties from 16% to 22% (+36%) by adding cross-border context. Public health modeling can now use monthly-refreshed, privacy-preserving location embeddings to reduce data-lag and help forecast and risk-predict with comparable or better accuracy versus conventional inputs.
Why it matters
Earth AI