In the suite of FAIR4AI projects, we are improving the practical usability of AI models by implementing the FAIR (Findable, Accessible, Interoperable, Reusable) principles. Currently, there are two projects addressing specific aspects of this technical and administrative part:
- FAIRmodels.org: Can we create a metadata repository for AI models in healthcare? This would improve how AI models are described and make it easier to package and reuse them in clinical practice.
- FAIVOR (FAIR AI validation and quality control): Can we build a system that retrieves models from the metadata repository and provides easy-to-use tools for external validation? This system would also enable reporting of validation results back into the FAIRmodels metadata repository.
