New Regulations Aim to Make AI Models More Useful, Less Biased
If you’re using patient electronic health information to develop algorithms for informing clinical care, you’ll want to take note of new rules published by the Office of the National Coordinator for Health Information (ONC).
With this new comprehensive document, titled “Health Data, Technology, and Interoperability: Certification Program Updates, Algorithm Transparency, and Information Sharing” (also known as the HTI-1 final rule), ONC is putting guardrails in place to help ensure we use patient data in ways that are responsible and safe. Most importantly, ONC wants to be sure algorithms using these data are fully transparent and lead to equitable and non-biased decisions about care.
Do these rules apply to you? You’ll particularly want to review the rules if you’re:
- developing software/algorithms for use with electronic health records (e.g., such as models that help inform patient care).
- working on solutions to make patient data more interoperable and secure.
- reporting metrics based on care delivery.
By implementing these provisions, you can help:
- make algorithms more transparent.
- facilitate access, sharing, and use of electronic health information.
- reduce the costs and work for health IT developers and people using health IT.
If you’re interested in reading about how the policies developed, the 800+ page document includes an extensive background section, along with comments from the public, and ONC’s responses to those comments.
Dr. Roxanne Jensen, program director of the Outcomes Research Branch, Healthcare Delivery Research Program, NCI’s Division of Cancer Control and Population Sciences (DCCPS), notes that these regulatory actions, which are in direct response to the 21st Century Cures Act, help bring greater transparency to the rapidly emerging area of AI-based predictive decision support tools. “My hope is that these guidelines will serve as benchmarks to foster increased awareness, comparability, and, ultimately, health equity in how we collect, use, share, and interpret patient data in both research and practice.”
For more information on ONC’s involvement in overseeing AI work, see our news article on leveraging health information technology, and read the new AI agreement between the United States and European Union.
To learn more about AI in cancer research, see:
- CBIIT’s Blogs, News items, and Funding Announcements and Notices related to AI.
- DCCPS’ Healthcare Delivery Research Program’s digital healthcare webpage, featuring NCI initiatives and research interests, along with current funding announcements.
- NCI’s new “AI in Cancer Research” webpage, with information on AI funding and events.