Health System AI Governance

As health systems implement artificial intelligence (AI) tools into their clinical care and operations, setting up appropriate governance is essential for ensuring safety and minimizing risk. Governance can provide a framework for health systems to be able to know what tools are being used within their organization, standardize their risk assessment and mitigation processes, and allow for documentation of processes. Governance frameworks generally include the clear articulation of principles and goals, predictability on what information will be needed as well as transparency on the process, identification of the participants involved in governance, and establish accountability and documentation.

Duke-Margolis researchers conducted a multi-part research project to better understand how health systems can effectively implement AI governance at their organizations. This project had three phases:

Health System Working Group: Duke-Margolis convened a working group comprised of health systems of varying sizes who have implemented their own AI governance processes, to share learnings and better understand considerations involved.

Expert Workshop: Duke-Margolis convened a multi-stakeholder expert workshop to diver deeper into this issue and to better understand the impact of health system AI governance on other stakeholders.

White Paper: After the expert workshop, the Duke-Margolis research team compiled a paper that explores the commonalities and differences in how different health systems are implementing AI governance, as well as considerations for other health systems who may be starting this process.