Planetary Prediction Engine automates geospatial data-to-model workflows

What changed. The system discovers signals, retrieves and fuses geospatial data and foundation embeddings, screens leakage, searches models, and applies overfitting guards. Reported results improve multiple US, Nigerian, and DRC benchmarks.

Technique / discovery. Modular agent orchestration, dynamic covariate selection, PDFM and AlphaEarth embeddings, leakage gates, and AutoML.

Why it matters. Scientific agents are moving into the expensive workflow between a question and a validated model, not only generating code or answers.

Application. Epidemiology, food security, disaster risk, environmental prediction, and spatial public health

Evidence maturity. Preprint (source tier A)

Identified bottleneck. Formal causal leakage verification, cross-scale noise, uncertainty, prospective validation, and generalization beyond one outbreak remain open.

Caveat / evidence note. Google Research arXiv preprint with author-reported comparisons. The outbreak result covers one pathogen and geography; operational decisions need external validation.

Review status. Reviewed. User requested: Yes.

• Google Research (opens in a new tab)

A reproduction tutorial is not yet available for this entry. The closest reviewed material is AI for science and Agent planning and cognitive architectures.