FAIRSCAPE: An AI-readiness Framework for Biomedical Research
Author Information
Author(s): Al Manir Sadnan, Levinson Maxwell Adam, Niestroy Justin, Churas Christopher, Parker Jillian A., Clark Timothy
Primary Institution: University of Virginia School of Medicine
Hypothesis
The FAIRSCAPE framework aims to enhance AI-readiness in biomedical datasets by providing comprehensive pre-model explainability.
Conclusion
The FAIRSCAPE framework successfully generates and integrates essential metadata for ethical and FAIR AI applications in biomedical research.
Supporting Evidence
- FAIRSCAPE provides ethical and semantic characterization of datasets.
- The framework integrates with NIH-recommended repositories.
- It supports twenty-two of the twenty-eight AI-readiness criteria.
Takeaway
FAIRSCAPE helps scientists make sure their data is ready for AI by providing clear information about how the data was collected and used.
Methodology
The framework uses Python packages to create, manage, and distribute Research Object (RO)-Crate packages with detailed provenance graphs.
Limitations
The framework currently supports only 22 of the 28 AI-readiness criteria recommended by the NIH Bridge2AI Program.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website