Health AI Partnership (HAIP) Community of Practice (CoP) Advocacy for Democratization of AI

The HAIP CoP has been at the forefront of advocating for policy orientations that ensure the democratization of AI technologies and capabilities. Our efforts are aimed at educating various stakeholders in AI lifecycle management, building local capacity, reducing the digital divide, validating AI systems responsibly, and ensuring equal access and use of AI.

Here is the summary of our efforts:

Events at the Hill Organized by HAIP

Landmark Summit on AI Product Lifecycle Management: In September 2024, HAIP organized a landmark summit titled “People, Process, and Technology for AI Product Lifecycle Management Within Healthcare Delivery Organizations” at the U.S. Capitol Visitor Center. This summit convened key stakeholders from congressional committees and federal agencies, fostering critical discussions on AI’s role in healthcare delivery.

Key Frameworks and Publications

HAIP Health Equity Across the AI Lifecycle (HEAAL) Framework: A community-informed framework developed through collaboration with an interdisciplinary team and inputs from 77 healthcare practitioners. It aims to mitigate the impacts of AI products on health inequities, focusing on five equity assessment domains: accountability, fairness, fitness for purpose, reliability and validity, and transparency. HEAAL is intended for use by healthcare delivery organizations considering AI solutions.

HAIP/DIHI Model Facts Label: Developed by the Duke Institute for Health Innovation (DIHI), first published in March 2020 in Nature Digital Medicine. Cited in the HTI-1 Final Rule, shaping federal AI healthcare regulations. The updated template, compliant with HTI-1, includes 31 source attributes and is available under Creative Commons Attribution 4.0 license for adaptation and distribution, with credit to DIHI.

Alignment with Standards: Healthcare organizations implementing AI solutions must comply with guidance from regulatory agencies, standards organizations, and oversight bodies. HAIP best practice guides, covering 8 key decision points in AI lifecycle management, map to prominent standards, aiding organizations in demonstrating alignment and enhancing governance. The Health AI Partnership offers tailored crosswalks for healthcare delivery organizations to meet these standards. Read the publication here

HAIP Vendor Disclosure Framework: A framework aimed at enhancing transparency and accountability among AI vendors and creating a standard procurement guide for health delivery organizations.

Advocacy and Policy Orientation

HAIP has contributed significantly to policy conversations through various publications and testimonies:

Eliminating the AI Digital Divide by Building Local Capacity: Provides an example of local capacity building using a hub-and-spoke network connecting health delivery organizations with technical, regulatory, and legal support services from vendors, law firms, and other health delivery organizations with more AI capabilities.

Editorial on Non-Discrimination Rules: Discusses challenges healthcare delivery organizations face in complying with non-discrimination rules related to AI.

Editorial on Equity Concerns: Highlights equity concerns related to centralized product testing at elite universities funded by big tech companies.

Alternate AI Regulation Approach: This proposes an approach inspired by the CLIA model and CMS rules governing laboratory tests.

FDA AI Guidance Concerns: Synthesizes concerns related to the FDA’s AI guidance published in 2022.

Engagement with Legislative Bodies

Testimony to US Senate Finance Committee: HAIP leadership council member Mark Sendak provided expert testimony on AI policy to the US Senate Finance Committee Hearing on AI in Health Care. [Written Remarks]

Congressional Briefing: Organized by the Alliance for Health Policy, the hearing offered valuable insights into AI in healthcare. Its goal was to establish a foundational dialogue for congressional staff and federal policymakers on how leaders in healthcare AI policy are developing and implementing standards for responsible AI use in the sector.

Multi-State AI Policy Working Group: The objective of the hearing was to gather feedback from participants on two pieces of AI legislation: the “High-Risk AI” framework and the Texas Responsible AI Governance Act. With representation from 47 states and over 200 lawmakers, the session aimed to address policy concerns, recommend changes, and compare the draft bills. Participants presented their recommendations and engaged in a Q&A session with lawmakers. [Overview here]

Health Information Technology Advisory Committee (HITAC) AI Public Hearing: The AI public hearing was organized by the Assistant Secretary for Technology Policy (ASTP), previously known as the Office of the National Coordinator for Health IT (ONC). See the HAIP members in the panel listed here. 

ARPA-H Site Visit and Panel: Leaders from ARPA-H attended an onsite event at Duke Health, facilitated by DIHI and Health AI Partnership, to explore AI and emerging technologies.

NC Local Legislatures Meeting: Briefed state and local legislatures Valerie Foushee (US Representative-NC District 4^th), Deborah Ross (US Representative-NC District 2^nd), and Zack Forde-Hawkins (NC Representatives-31^st district) on AI deployment and governance in healthcare, and the work of HAIP in surfacing best practices for AI lifecycle management.

EPIC Wisconsin Roundtable: Participated in a roundtable with 25 national leaders to discuss AI in healthcare. Review event website and report.

HHS Deputy Secretary Visit at Duke: Leaders from the U.S. Department of Health and Human Services, Microsoft, and UNC-Chapel Hill visited Duke to discuss the responsible use of AI in healthcare, focusing on opportunities, risks, and strategies to ensure trustworthy AI implementation.

Contributions to HTI-2 Taskforce: contributed significantly to the HTI-2 Final Rule, which improves electronic health information exchange by enhancing interoperability, privacy, security, and trust within TEFCA, and updating information blocking regulations. (No longer active under the current administration).

Mentions in Regulatory/Policy Documents

Cited in the Federal Legislation on AI Transparency: HAIP’s founding team at Duke Institute for Health Innovation was the first to publish the Model Facts Label for AI products in healthcare.  Read the final rule on Health Data, Technology, and Interoperability: Certification Program Updates, Algorithm Transparency, and Information Sharing (HTI-1) Final Rule citing the use of Model Fact Label.

Pilots for Capacity Building

HAIP Practice Network Program: This is the first technical assistance program that supports safety net organizations (SNOs) (i.e., Federally Qualified Health Centers (FQHCs) and community hospitals) adopting HAIP’s AI best practices. The 2024-2025 cohort receives implementation guidance, expert consultation, peer learning, educational materials, and resources. These organizations will share learnings and promote safe, effective, and equitable AI use in healthcare.

HAIP Practice Network in California: The Health AI Partnership (HAIP), in partnership with The SCAN Foundation and California Health Care Foundation (CHCF), is developing a coordinated technical assistance program to support the effective implementation of artificial intelligence (AI) among California’s safety-net providers. By fostering connections across the healthcare ecosystem and promoting access to expert guidance and support, the program seeks to close the digital divide in healthcare and ensure all providers–regardless of size or resources–can evaluate, implement, and monitor AI technologies responsibly and effectively.