Google Earth AI is a suite of geospatial models and data that can help public-health teams estimate local needs and forecast risks when health records are sparse, delayed, or difficult to compare across administrative boundaries. Its Population Dynamics Foundation Model (PDFM) supplies location-level signals for other models; it does not diagnose people or independently prevent outbreaks. Google’s October 6, 2026 account reports partner evaluations across several tasks, but their results are specific to the disease, geography, metric, and baseline tested.
What Google Earth AI does for public health
Google describes Earth AI as a geospatial model and data suite that combines satellite imagery, weather, aggregated search trends, mobility, and other population-dynamics signals. One component, the Population Dynamics Foundation Model (PDFM), uses self-supervised learning to turn privacy-preserving signals into compact location embeddings—described by Google as “fingerprints” for places. The embeddings are refreshed monthly and can be added to existing machine-learning workflows.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
MOVA Self Rotating Globe, Ocean Blue Relief Classic (4.5") | $275.00 | Buy on Amazon |
| 2 |
|
MOKOQI Magnetic Levitating Globe with LED Light, Cool Tech Gift for Men Father Boys and Girls,... | $41.99 | Buy on Amazon |
As an Amazon Associate I earn from qualifying purchases.
Signals described for PDFM include aggregated frequencies of search topics; local density and busyness around places such as pharmacies, clinics, and parks; and high-resolution weather and air-quality measures. These are indicators about locations and their environments, not individual health records. A public-health team can combine them with its own health data and conventional inputs, then use a separate epidemiological or statistical model to estimate a relevant outcome.
This distinction matters when interpreting a forecast: PDFM is an input, not a standalone disease-prediction system. Google says its signals may complement local records where those records are limited, out of date, or constrained by administrative boundaries. Whether the added signal helps depends on the task, local data, and how the model is validated.
#1 Best Overall
- Rotates Silently Using Light, No Batteries Needed - MOVA Globes spin continuously using ambient indoor light and the Earth’s magnetic field. No cords. No noise. Just smooth, self-sustained motion.
- Detailed Terrain Relief Design - Classic blue ocean map design features raised topography and realistic terrain colors—perfect for geography lovers, educators, and world travelers.
- Premium Build with Seamless Movement - A fluid-suspended inner globe rotates within a clear acrylic outer shell, offering a smooth, almost magical visual experience.
- Compact Yet Eye-Catching - Measures 4.5 inches in diameter and stands about 7 inches tall with the included clear acrylic stand—perfect for desks, bookshelves, or reception counters.
- Ideal Gift for Learners, Thinkers, and Explorers - Whether for a classroom, office, or thoughtful gift, this modern interpretation of the traditional world globe brings both beauty and quiet motion to any space.
What partner evaluations have reported
Google Research’s October 6, 2026 overview lists five partner evaluations. The results below are not one general accuracy score: each concerns a different outcome, geographic unit, baseline, and metric.
| Task and setting | Reported result | How to interpret it |
|---|---|---|
| Cross-border measles, mumps, and rubella (MMR) vaccination modeling in 146 U.S. counties near Canada | Adding Canadian context raised explained variance from 16% to 22%, a 36% relative gain. Google Research reports that estimates refined coverage by at least three percentage points for 4.7 million border residents. | This is evidence that information across a border helped refine coverage estimates in this setting. It is not a reported change in vaccination rates or health outcomes. |
| Cardiovascular-mortality nowcasting across 3,091 contiguous-U.S. counties | PDFM-based models had a mean absolute error of 18.7 deaths per county, compared with 19.1 for census-based models; RMSE was 46.00 versus 57.69. Google says the differences were not statistically significant. | The results suggest PDFM could be a timely stand-in when census data are stale or unavailable, not a statistically established accuracy advantage over the census-based models. |
| Dengue forecasting in roughly 2,450 Mexican municipalities | With TimesFM integration, the evaluation reported a statistically significant one-month gain: weighted interval score difference −0.0051, with lower scores better. Accuracy improved in up to 72% of municipalities with active transmission. | The gain concerns the evaluated forecasting setup and municipalities with active transmission; it does not establish the same improvement in other countries or at other forecast horizons. |
| Postpartum-depression risk prediction using 332,970 U.S. CDC PRAMS respondents | The reported AUC gain was 0.0020 over a 0.62 baseline in states represented in training and 0.0038 in unseen states. Google says area poverty was encoded at R² = 0.45 and the embedding recovered about 15% of the predictive signal of income and insurance records. | This is a small improvement in a risk-prediction metric, not a replacement for individual socioeconomic information or a solution to screening gaps. |
| Cholera emergence in 403 health zones in the Democratic Republic of the Congo over 89 weeks | Google Research reports statistically significant relative gains of 9.7% in area under the precision-recall curve at four weeks and 18.1% in Precision@5 at eight weeks; the Precision@5 gain was 19.3% in endemic zones. | Precision@5 measures how often the model’s five highest-risk zones in a given week later had an outbreak. It ranks zones for attention; a high-risk signal is not confirmation that an outbreak has begun. |
These figures are findings reported in Google Research’s October 6, 2026 overview. The available account does not establish that every result has been independently replicated or that the performance transfers to other populations, places, or operating conditions.
How the approach has been used beyond those evaluations
Ebola response mapping and forecasts in the DRC
In a separate October 6, 2026 account, Google describes work with WHO AFRO and the DRC’s INRB during an Ebola response. Google says a prototype Geospatial Reasoning agent mapped 48 exposed settlements and more than 45,500 people considered at risk along remote mining corridors in minutes, compared with a process Google says would normally take weeks. The account also describes weekly spread-risk forecasts for uninfected zones using mobility flows, historical case trends, and Earth AI models. These are Google’s descriptions of operational work, not independently verified performance findings here; a risk forecast is not a confirmed case or outbreak.
Earlier projects reported in March 2026
Google’s March 13, 2026 overview describes a Cooper/Smith project combining PDFM and AlphaEarth satellite embeddings with local data to predict clinic utilization in Malawi; ZIP-code-level estimates of measles vaccination coverage; cholera forecasting with TimesFM, PDFM, and weather data, reported as more than 35% more accurate than standard models; dengue forecasting in Brazil; and an Australia proof of concept for chronic-disease needs. These are separate examples from the October evaluations, with different studies and metrics. The reported “over 35%” should not be treated as a summary score for Earth AI or combined with the later cholera findings.
What organizations can access
Google says PDFM embeddings are commercially available in Preview as Population Dynamics Insights, a geospatial embeddings dataset from Google Maps Platform. Researchers can request no-cost access for select use cases, and eligible organizations can apply for Google Earth credits through Google Maps Platform Public Programs. Access is conditional: the October 6, 2026 account does not specify universal eligibility, prices, service commitments, or processing timelines.
Rank #2
- UPGRADE LEVITATION GLOBE: Globe floating in midair by the magnetic force from bottom and top of frame. LED beads in frame and Globe shows Red,Orange,Yellow,Green,Blue,Purple, Cyan 7 colors. Amazing centerpiece
- AUTOMATIC SPINNING: Floating Globe Levitating Globes- Almost zero friction force while levitating in the air. Rotation drives in the globe, the 360-degree rotation all the time. The cool decor on a desk or Book Shelf Decor. Cool conversation piece with others
- COOL GIFT for MEN & KIDS: Great gifts for boyfriend, children, colleagues, even your boss. Perfect birthday gifts for men and personalized Christmas Gifts for men, Anniversary or Valentines day gifts, Father's Days gift ideas. It is a funny men’s gifts
- EASY INSTALL: Globe diameter 3.5 inch /9 cm, easily hold by 1 hand. Switch on the C shape frame, Squat down your body and ensure sight line height is same to the opening of frame. Then follow the user manual and you will success after several times try
- WHAT YOU GET: Stand, Globe ball, Power adapter. Globe ball shows country name, ocean, continent by clear words. Attract boys / girls and inspire their exploring spirit and interesting to geography and STEM science
The Geospatial Reasoning agent and planetary prediction engine described in the Ebola response are research prototypes, not the same offering as the Population Dynamics Insights Preview. The public account does not establish that these prototypes are generally available to organizations.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What to validate before using geospatial predictions
A useful evaluation starts with the decision a health team needs to make—not with a model’s broad label. Nowcasting, short-horizon outbreak forecasting, risk stratification, and mapping exposed settlements have different targets and consequences. Before using an output to direct staff or resources, a team should assess:
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Local fit: Check whether the people, places, time periods, and disease patterns represented in the model match the population and setting where it will be used.
- Baseline and metric: Compare against a relevant local or conventional approach, and choose measures that match the decision. A ranking metric, an error measure, and explained variance answer different questions.
- Timing and data coverage: Establish how current the inputs are and whether important communities or locations are underrepresented. Monthly embeddings do not by themselves guarantee that every underlying signal is equally current.
- Human review and response capacity: Decide who interprets risk signals, how they are checked against local surveillance, and whether the health system can act on them. A forecast cannot replace confirmation, clinical judgment, or response resources.
- Accountability and equity: Set rules for oversight, explain how model outputs affect decisions, and monitor whether benefits or burdens differ across communities.
The World Health Organization’s guidance on AI for health emphasizes evidence-based adoption, governance, ethics, trust, safety, equity, and sustainable implementation across health systems. Those considerations apply alongside predictive performance: a technically useful signal is not sufficient if local teams cannot validate it, access it, or respond fairly.
The Google accounts characterize PDFM as privacy-preserving, but the information available here does not independently assess its privacy architecture, fairness across populations, performance under all deployment conditions, or whether earlier predictions produced measured health outcomes. Those questions require evidence specific to the system and setting in which it would be used.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
