Authors: Amelia Jiménez-Sánchez, Natalia-Rozalia Avlona, Dovile Juodelyte, Théo Sourget, Caroline Vang-Larsen, Hubert Dariusz Zając, Veronika Cheplygina

Published on: February 09, 2024

Impact Score: 8.12

Arxiv code: Arxiv:2402.06353

Summary

• What is new: Introduces ‘actionability’ as a metric to evaluate the quality gap in medical imaging datasets on Community-Contributed Platforms (CCPs).
• Why this is important: The current governance model of CCPs fails to maintain the necessary quality and recommended practices for medical imaging datasets, impacting the effectiveness of AI in healthcare.
• What the research proposes: A commons-based stewardship model for improving the documentation, sharing, and maintenance of datasets on CCPs.
• Results: Analysis of 20 datasets reveals significant issues like vague licenses, lack of identifiers, duplicates, and missing metadata. The proposed stewardship model addresses these issues.

Technical Details

Technological frameworks used: nan

Models used: nan

Data used: 20 medical and computer vision datasets from CCPs

Potential Impact

Medical imaging, AI healthcare companies, and Community-Contributed Platforms (e.g., Kaggle, HuggingFace) could benefit or need to adapt based on these insights.

Want to implement this idea in a business?

We have generated a startup concept here: MediDataGuard.