Mantis: A Simulation-Grounded Foundation Model for Disease Forecasting

2025-08-17

AI Summary

Traditional epidemiological forecasting models rely on disease-specific surveillance data and expert-tuned parameters, resulting in poor generalizability to emerging outbreaks and resource-limited settings. To address this, we propose the first purely mechanism-driven foundational model for infectious disease forecasting—eliminating all dependence on real-world case data. Our approach leverages over 400 million simulated outbreak-days, encompassing diverse pathogens, transmission routes, intervention policies, and surveillance noise, to train a generalizable predictive framework. The model supports interpretable, eight-week-ahead forecasts and transfers effectively across diseases, geographies, and transmission dynamics. Evaluated on six distinct infectious diseases, it significantly outperforms 39 expert-optimized baselines—including all models deployed by the CDC’s COVID-19 Forecasting Hub—despite being trained exclusively on synthetic data. This demonstrates high-accuracy, long-horizon forecasting without any real-world incidence data.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Machine Learning: Other Foundations of Machine Learning
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Infectious disease forecasting in novel outbreaks or low resource settings has been limited by the need for disease-specific data, bespoke training, and expert tuning. We introduce Mantis, a foundation model trained entirely on mechanistic simulations, which enables out-of-the-box forecasting across diseases, regions, and outcomes, even in settings with limited historical data. Mantis is built on over 400 million simulated days of outbreak dynamics spanning diverse pathogens, transmission modes, interventions, and surveillance artifacts. Despite requiring no real-world data during training, Mantis outperformed 39 expert-tuned models we tested across six diseases, including all models in the CDC's COVID-19 Forecast Hub. Mantis generalized to novel epidemiological regimes, including diseases with held-out transmission mechanisms, demonstrating that it captures fundamental contagion dynamics. Critically, Mantis is mechanistically interpretable, enabling public health decision-makers to identify the latent drivers behind its predictions. Finally, Mantis delivers accurate forecasts at 8-week horizons, more than doubling the actionable range of most models, enabling proactive public health planning. Together, these capabilities position Mantis as a foundation for next-generation disease forecasting systems: general, interpretable, and deployable where traditional models fail.

Problem

• Enables disease forecasting without disease-specific data
• Generalizes across diseases and regions with limited data
• Provides interpretable forecasts for public health decisions

Innovation

• Trained on 400M simulated outbreak days
• Outperforms 39 expert-tuned models
• Mechanistically interpretable for decision-makers