In a first for the U.S. health care industry, physicians and administrators in health care delivery settings have come together to create and disseminate continually updated best practices for implementing augmented intelligence (AI)—often called artificial intelligence—and to cultivate a community of practice.
The Health AI Partnership is a collaboration among 14 health care organizations and ecosystem partners, including Duke Health, Mayo Clinic, the University of California at Berkeley, the AMA, and DLA Piper, a global law firm.
The keystone in that effort is the Health AI Partnership website, which features a collection of topic guides for clinicians and health care leaders looking to adopt AI. The partnership is seeking feedback on the website’s content by June 16.
“There are 7,000 hospitals and hundreds of thousands of front-line clinicians in the U.S., and many are simply unable to keep pace with changes in technology and the regulatory guardrails,” said Mark Sendak, MD, the population health and data science lead at the Duke Institute for Health Innovation. “We want to hear if clinicians have developed their own strategies for tackling these problems. This is meant to be a communal resource with community-generated insights.”
Learn more about augmented intelligence and the AMA’s other research and advocacy in this vital and emerging area of medical innovation.
The website’s topic guides explore key decision points in all four phases of implementing an AI product—procurement, development, integration and lifecycle management.
For example, within procurement, users are encouraged to first identify and prioritize a problem. Individual topic guides then help users identify problems across the organization, prioritize problems, determine potential downstream impacts, ascertain the dimensions of the problem and evaluate the suitability of technical approaches.
Each topic guide begins by answering the question of why the issue matters and then explains, step by step, how to do the work.
“We built this so that regardless of how much you know about AI, and whether you're looking to purchase something or you already use something in practice, there is guidance relevant to you,” Dr. Sendak said.
“This is not meant to be something that you have to spend hours digging through,” he added. “It’s meant to be rapid so you can pop in and get some information about a challenge you're facing. Then maybe a couple weeks later you can pop in and get something totally different.”
To encourage the spread of best practices, the Health AI Partnership—funded by the Gordon and Betty Moore Foundation—embraces these four values.
Advancing equity. The partnership features organizations that focus on safety-net contexts, and all of its guides are available free. In addition, the content covers AI products developed both in-house and by external vendors.
Improving patient care. The website always frames AI as just one component of a solution, which might also include clinical workflows, organizational change management, education campaigns and feedback loops to clinical end-users and patients. It focuses on AI tools that assist with diagnosis or treatment decisions and that help prioritize patients when clinical resources are scarce.
Improving the clinical work environment. The topic guides advocate involving front-line clinicians throughout the lifecycle to ensure the use of AI alleviates the burdens they face. Moreover, AI is not framed as a substitute for human labor but rather a way to improve the safety and efficiency of care.
Facilitating participation. The content will be refreshed at regular intervals. Case studies are being solicited from collaborating organizations, and health care delivery leaders are invited to help create and refine new topic guides.
“We have intentionally created an environment that is focused on the delivery system,” Dr. Sendak said. “We’re not prioritizing engagement with vendors. We’re focused on the concerns of clinicians.”
Learn more about the AMA's commitment to helping physicians harness health care AI in ways that safely and effectively improve patient care.