Scientific prototype for systemic pandemic strategies powered by AI
In the COCAP project (short for COping CAPacity of nations facing systemic crisis -a global intercomparison exploring the SARS-CoV-2 pandemic), one of the main goals is to develop a first scientific prototype of a decision support system (DSS), which is planned to be setup in the beginning of 2026. This system is designed to recommend a systemically optimal set of non-pharmaceutical interventions (NPIs) – such as mask mandates or limits on public gatherings – tailored to both the nature of a specific pathogen and the national context.
The DSS is built as a modular framework, meaning it can flexibly combine advanced models. These models simulate how infectious diseases spread and assess the economic impact of each measure. Also, other relevant indicators and corresponding models, such as mobility or quality of life, can be included. This kind of system is seen as a key tool for strengthening systemic resilience against pandemics.
A major strength of the DSS is its adaptability. Each model is carefully calibrated with input from experts to ensure it matches real-world data. To reflect the uncertainty of epidemic dynamics, the system includes a range of possible outcomes – based on both the characteristics of the pathogen and human behavior. This allows users to explore worst-case and moderate scenarios in one simulation.
Users or decision makers can build and explore a wide variety of scenarios by selecting different pathogens, countries, or available NPIs. Based on the user’s priorities – such as the relative importance of health, economy, or mobility – the system will suggest a dynamic and systemically balanced set of NPIs to apply at specific times. These recommendations can be developed manually or generated by AI agents that can handle the vast number of possible intervention combinations, while keeping the results practical, understandable, and aligned with user-defined preferences.
At its core, the system combines agent-based simulations and differential equations with AI-driven optimization. This ensures that recommendations are grounded in scientific modeling and enhanced by AI’s ability to learn and make decisions. This integration allows the DSS to deliver actionable insights, even in complex and uncertain situations.
In summary, this first generic and customizable scientific demonstrator for integrated pandemic management offers a method for designing tailored, system-level intervention strategies for nearly any endemic or pandemic scenario. By combining cutting-edge models, expert calibration, and AI optimization, it provides robust, evidence-based recommendations that balance disease control with broader societal impacts.
Associated institute at KIT: Institute for Thermal Energy Technology and Safety (ITES) – Department Resilient and Smart Infrastructure Systems (RESIS)
Authors: Sadeeb Simon Ottenburger, Raphael Dujadin
