Roadmap: An Ontology of Medical AI Models and Datasets.
Authors
Affiliations (6)
Affiliations (6)
- David Geffen School of Medicine, University of California, Los Angeles, Calif.
- Department of Radiology, University of North Carolina, Chapel Hill, NC.
- Department of Radiology, University of California, San Diego, Calif.
- Balliol College, Radcliffe Department of Medicine, Oxford University, Oxford, UK.
- Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Republic of Korea.
- Department of Radiology and Institute for Biomedical Informatics, Perelman School of Medicine, University of Pennsylvania, 3400 Spruce St, 1 Silverstein, Philadelphia, PA 19104-6243.
Abstract
Successful development, regulatory review, and clinical implementation of artificial intelligence (AI) systems in medicine require clear, unambiguous communication about AI models and datasets. The Radiology Ontology of AI Datasets, Models, and Projects (ROADMAP) was developed to provide a machine-interpretable framework to describe medical AI resources by formally defining the attributes of AI models and datasets and their allowable values. ROADMAP builds upon generalized "model cards" and "datasheets for datasets" by incorporating features that support multimodal data, including medical images, structured data, and unstructured text. ROADMAP references concepts from widely used ontologies, coding schemes, and common data elements to improve the discoverability, interoperability, and reuse of AI resources. The ontology can facilitate matching of appropriate AI models with relevant datasets and support the detection of potential sources of bias in AI resources; it is available at https://bioportal.bioontology.org/ontologies/ROADMAP. © RSNA, 2026 See also accompanying Special Report on ROADMAP and metrics.