In an era defined by unprecedented global interconnectedness and shifting environmental patterns, public health needs a new generation of tools to anticipate and manage health risks. A virus emerging in a remote village can traverse continents in a single day, while shifting weather patterns expand the habitats of disease-carrying vectors. Yet, our current global surveillance systems remain reactive and fragmented. We are fighting 21st-century biological threats with 20th-century tools, leading to preventable mortality and significant economic strain. To build true health security, we will have to pivot from passive observation to proactive intelligence, a shift made possible by the convergence of geospatial data and artificial intelligence. 

From observation to proactive intelligence 

Current health surveillance efforts are often constrained by fragmented data ecosystems. Epidemiological data often sits isolated from the environmental contexts that drive disease transmission. Geospatial initiatives, such as Earth AI, bridge this gap by integrating global datasets along with foundational models of the planet—weather, floods, and population dynamics—with agentic reasoning capabilities. This approach acknowledges that public health is not just about biology; it is about the multiple aspects that influence health outcomes including geography, climate, and economic factors. 

The power of this technology lies in its ability to transform raw, disparate data into actionable intelligence. For instance, consider the challenge of resource allocation in resource-constrained environments. In Malawi, the integration of a Population Dynamics Foundation Model (PDFM) with satellite embeddings and mobile call records outperformed traditional geostatistical and deterministic models for 87% of health care indicators. These insights enable health leaders to transition from reactive reporting to proactive planning, by predicting disease outbreaks and optimizing the allocation of critical resources like vaccines and medication. Similarly, in Brazil, by integrating the PDFM model with TimesFM, researchers enhanced long-term forecasting reliability for dengue, achieving a 44% increase in predictive accuracy for 12-month projections.¹ This advancement provides a more stable foundation for long-range planning, directly addressing the historical difficulty of maintaining model performance over extended time horizons. 

By integrating nontraditional data streams such as satellite imagery, mobile phone records, and search data, embeddings offer an efficient means of supplementing existing health information systems. While they cannot replace robust primary data collection, embeddings may provide a valuable tool for identifying trends, filling geographic gaps, and informing resource allocation in real time.

Answering complex "what if" scenarios 

However, the true revolution of geospatial insights lies beyond targeted prediction by enabling dynamic agentic reasoning. We are moving toward systems capable of answering complex "what if" scenarios. Imagine a "Geospatial Reasoning Agent" that acts as a central coordinator for disease surveillance. A public health official could ask, "What happens to the forecast for cholera in this district if we distribute 10,000 oral rehydration salts today?" or "How will the predicted cyclone impact pharmacy access in the next 48 hours?". The agent has the ability to automatically pick the best datasets, consult state-of-the-art foundation models, and assemble them to build a comprehensive answer that accelerates the ability of public health officials to have timely and actionable insights.  

This capability offers a transformative shift for crisis response, recognizing that health outcomes are shaped by the interaction between biology, the environment, and circumstances in which people live. By layering real-time environmental data over health data and community vulnerability metrics, geospatial reasoning allows us to predict crises before they manifest. Outbreaks disrupt global supply chains, strain public health workforce and healthcare facilities, and widen health inequities. A proactive, intelligence-driven approach offers direct benefits by minimizing the impact of events and improving health outcomes.

Global health security and the complex interplay between the planet and its people

Ultimately, global health security depends on our ability to understand the complex interplay between the planet and its people. Geospatial reasoning tools offer planet-scale data and reasoning capabilities necessary to make sense of this complexity. By using advanced artificial intelligence tools, we can improve data ingestion, risk modeling, and outcome measurement. With the technical foundation in place, the opportunity now is to thoughtfully integrate these insights into the existing strategies of public health leaders dedicated to promoting well-being for all. 

The views expressed in this commentary are those of the author and do not necessarily reflect the views of the World Health Summit (WHS). This commentary is part of the ongoing WHS Perspectives series on health security. Additional perspectives from other contributors have been and will be published in the coming weeks.