Partnerships
Coalition for Health AI (CHAI)
Duke Health is proud to be a founding member of the Coalition for Health AI (CHAI). As the health AI landscape continues to evolve, Duke’s AI Evaluation & Governance leadership remains committed to fostering responsible and trustworthy innovation while aligning with CHAI’s principles and guidance.
CHAI is a multi-organizational partnership dedicated to advancing best practices for AI in healthcare. Since its inception, CHAI has grown into a dynamic non-profit organization, building on the foundational work of expert volunteers to establish consensus-driven guidelines for health AI. By collaborating with a diverse network of members, including health systems, research institutions, technology developers, payers, and patient advocacy groups, CHAI is shaping the future of trustworthy AI.
In April 2023, CHAI released the Blueprint for Trustworthy AI, developed with input from stakeholders across healthcare, academia, government, and industry. This guidance reflects a broad range of expertise and perspectives, setting foundational principles for best practices in the responsible development, deployment, and use of health AI.
Workgroups developed the consensus-driven framework around CHAI’s key principles: Usefulness, Fairness, Safety, Transparency, and Security & Privacy. Contributors included professionals from medicine, data science, technology development, patient advocacy, and academia — representing institutions such as Historically Black Colleges and Universities (HBCUs), major and emerging tech companies, and government agencies like the FDA. This process culminated in the creation of CHAI’s Responsible AI Guide and the accompanying reporting checklists.
Duke Health's AI Evaluation & Governance team was pivotal in CHAI's inception, shaping its vision, structure, and foundational principles. From convening early collaborators to providing essential scientific and operational expertise, Duke transformed CHAI from a voluntary effort into a leading non-profit championing trustworthy and responsible AI. Duke Health also formalized CHAI's consensus-driven approach and supported the development and dissemination of its key frameworks, contributing significantly to CHAI's growth. As a CHAI member, Duke continues to provide critical thought leadership and expertise, reinforcing CHAI's mission to establish trust, accountability, and best practices for responsible health AI.
Trustworthy & Responsible AI Network (TRAIN)TM
As a founding member of the Trustworthy & Responsible AI Network (TRAIN), Duke Health is dedicated to strategically operationalizing and democratizing trustworthy and responsible AI. TRAIN’s members are committed to sharing insights on learnings from procuring, adopting, developing, and using AI solutions amongst its member healthcare organizations. The network’s overarching goal is to ensure the responsible and trustworthy use of AI in healthcare, thereby enabling effective, ethical, and efficient AI-driven healthcare for the benefit of all.
At the forefront of health AI collaboration, TRAIN translates responsible AI principles into practical, actionable strategies. It focuses on real-world applications, emphasizing value assessment, outcome measurement, and knowledge exchange among its members. By prioritizing safety, quality, and trust, TRAIN is setting new standards for AI-driven healthcare.
As the practical implementation arm complementing the best practice development, policy, and advocacy efforts of the Coalition for Health AI (CHAI), TRAIN provides healthcare organizations with the tools, frameworks, and structured approaches needed to integrate AI responsibly into clinical workflows.
Duke Health plays a central role in shaping and advancing TRAIN’s mission. As a founding and highly active partner, Duke contributes expertise and resources to ensure AI systems are rigorously tested, validated, and continuously monitored for reliability and security. Duke Health’s leadership helps extend the benefits of AI across diverse healthcare settings, addressing disparities and promoting fairness in AI-driven decision-making.
Collaborations
Duke Health-VUMC Partnership and the Maturity Model Project
Recognizing a significant gap in the current oversight, resources, and organizational infrastructure needed to responsibly manage algorithmic technologies in health systems, the Health AI Readiness project focuses on addressing challenges in documentation, accountability, and governance of AI models.
In 2023, Duke University School of Medicine and the Vanderbilt University Medical Center (VUMC) were awarded a $1.25 million grant from the Gordon and Betty Moore Foundation for the project “Measuring Artificial Intelligence (AI) Maturity in Healthcare Organizations.” The Duke-VUMC Maturity Model Project aims to create a framework that identifies and establishes the essential capabilities health systems must develop to ensure the trustworthy and ethical utilization of AI models in healthcare.
At present, the proposed maturity model framework is being developed as a practical tool for health systems, allowing them to better assess, document, and improve their readiness for implementing AI technologies. The framework emphasizes essential elements such as technological infrastructure, data quality, governance and oversight, ethical considerations, and workforce competencies—all aimed at enhancing the safety, fairness, and quality of AI in health care.
The project teams are engaging with a wide range of stakeholders, including clinicians, administrators, policymakers, and technical experts, to determine the significant components necessary for the trustworthy implementation of AI in healthcare. These engagements will inform the development of a structured framework that health systems can use to benchmark their progress, identify gaps, and implement improvements to responsibly deploy AI solutions. The initiative will also consider the evolving regulatory landscape and address specific risks and challenges associated with AI in healthcare, such as algorithmic bias, transparency, and equitable access to AI-driven benefits.
Led by experts like Michael Pencina and Nicoleta Economou from Duke and Peter Embí and Laurie Novak from VUMC, the project team is drawing upon a diverse array of experiences and expertise in clinical care, data science, AI ethics, and health system management. Their collective vision is to transform healthcare through the responsible and impactful integration of AI, promoting a future where AI-driven innovations support better patient outcomes while upholding the highest standards of safety, fairness, and ethical integrity.
The outcomes of this project are expected to provide a blueprint for health systems across the nation, guiding them toward an AI-augmented future that is both effective and accountable to the patients and communities they serve.