AI-Generated Digital Twin Infrastructure for Healthcare & Living Environment

Digital twins, virtual replicas of patients, buildings or whole neighbourhoods, are among the most powerful tools available for testing decisions before they are taken in the real world. Today, building one still demands scarce expertise, bespoke code and months of integration work, which keeps the technology out of reach for most clinicians, researchers, housing operators and policymakers who could benefit from it.

AIGENT (AI-Generated Digital Twin Infrastructure for Healthcare & Living Environment) sets out to change that. Coordinated by VITO and bringing together 16 partners from nine countries, the project is building a pan-European infrastructure that lets users design, deploy and scale trustworthy digital twins, in healthcare, in the living environment and in other scientific fields, without writing code.

The infrastructure rests on a three-tier fabric: Edge Twin Nodes (ETNs) for privacy-preserving local processing, the AIGENT Nexus (ANX) for orchestration and knowledge management, and the AIGENT Compute Platform (ACP), which taps EuroHPC and the European AI Factories for large-scale training and simulation. Generative AI sits on top as a no-code “Scenario Studio”, where a user describes a problem in plain language and the platform assembles and launches the corresponding simulation.

The project will:

• Deliver an open, no-code GenAI Twin-Builder Platform that configures and orchestrates simulations and AI tools across research infrastructures.
• Validate the approach in six pilots in Belgium, Finland and Greece, three in healthcare (cardio-respiratory health, adverse drug reactions in older adults, adaptive healthy living) and three in the living environment (building energy optimisation, district fleet analytics, indoor environmental quality), each reaching TRL 6-7.
• Put decision-makers in the driving seat through visual “scenario cards”, with a target of at least 70% of pilot scenarios configured by non-technical users.
• Build trust into the stack by design, with explainability, uncertainty quantification, signed and reproducible run manifests, and governance aligned with the AI Act, GDPR, the EHDS and the EPBD.
• Share what works through a Twin Adoption & Co-Design Playbook, VVUQ guidance, privacy-preserving synthetic datasets and standards-ready assets (ISO, CEN, HL7, OGC).

AIGENT is deliberately built on mature, widely used components and open standards, and aligns with AI-on-Demand, GenAI4EU, EuroHPC AI Factories and the Virtual Human Twin initiative. The aim is reusable public infrastructure: faster deployment, lower integration costs and decision-ready insight for the people who need it, from GPs and older citizens to building operators and policymakers.