AdvaMed Digital Health Board Approves Voluntary Data Access Best Practices
The trade group's framework sets four principles for how device makers and U.S. health systems share, protect and use data, including for AI tools.

Key points
- AdvaMed said on October 28 that its Digital Health Tech Board of Directors approved voluntary Data Access Principles & Best Practices.
- The framework is built on four principles: transparency, data stewardship, de-identification and responsible use of AI and advanced analytics.
- AdvaMed said the best practices aim to reduce contracting and implementation delays between medtech companies and health systems.
AdvaMed, the medical technology trade association, said on October 28 that the board of its Digital Health Tech division has approved a set of voluntary principles and best practices for how device and software makers access, use and protect data when they work with U.S. health systems.
According to AdvaMed's announcement, the document is formally titled AdvaMed Data Access Principles & Best Practices for Introducing Technology into a Health System. The group said it is meant to help companies and health systems "collaborate on responsible data access, use, and protection," with patient trust at the center of efforts to improve care, support research and advance innovation, including artificial intelligence.
Four principles
AdvaMed said the framework sets common expectations for the data needed to implement a technology, fit it into clinical workflows, optimize its performance, monitor safety and quality, and improve products and services. It covers both health information protected by federal health privacy law and other clinical, device and operational data.
The best practices are organized around four principles, according to the release:
- Transparency and clear expectations: explaining what data are collected, how they will be used and protected, and each party's responsibilities, through agreements, documentation and privacy disclosures.
- Responsible data stewardship: following applicable laws, limiting identifiable data to permissible purposes, addressing retention, and supporting accountability through governance, training and oversight.
- De-identification as a foundation for innovation: using recognized de-identification methods, regularly assessing how well they work, and applying safeguards to reduce the risk that individuals could be identified again.
- Responsible use of AI and advanced analytics: transparent data practices and risk-based oversight, including evaluation of performance, safety and potential bias across a technology's lifecycle.
The framework also supports two-way sharing, the group said, including safeguarded benchmarking, quality and safety information, and real-world data that health systems can use. AdvaMed said the practices "aim to reduce contracting and implementation delays while supporting productive, trusted partnerships." The principles are voluntary; the release does not describe any enforcement or certification process.
What industry leaders said
Robert Cohen, chair of the Digital Health Tech board and vice president of innovation and technology in Stryker's orthopaedic group, said in the release: "High-quality data help us develop, validate, and improve AI-enabled tools." He said companies may need appropriate access to information about what happens before, during and after a procedure, not only in the operating room, to optimize a surgical technology's performance.
Mick Farrell, chair of AdvaMed's board and chairman and CEO of Resmed, said: "Responsible access to data can help us personalize each health journey and learn how technology performs in the real world."
AdvaMed president and CEO Scott Whitaker said: "Protecting patient privacy and advancing innovation must go hand in hand." He said the practices give companies and health systems "a practical way to build trust, work through data-sharing questions, and keep the focus where it belongs: better care for patients."
Part of a broader AI push
The release came the same week AdvaMed published comments to the Food and Drug Administration on its discussion paper about generative AI-enabled medical devices, which the group's site lists as a separate resource dated October 30. AdvaMed's site also lists the data access framework itself as a resource dated October 27.