AI’s dramatic impact has become a central topic for leaders across the health care industry. While AI solutions are not new to health care, the newfound accessibility of generative AI models raises a host of strategic, practical—and at times, uncomfortable—questions. As leaders navigate an emerging AI-powered future, are they ready to address common misconceptions, push beyond the headlines, and lead their organizations through the acceleration of AI in health care?
At 2023 Health Evolution Connect, leaders from across the industry came together to discuss the challenges and opportunities for executives as they look to navigate and harness the power of AI. In a discussion titled “Navigating an AI-Powered Future: Preparing Leaders from the Boardroom to the Frontlines for the Growth of AI in Health Care,” moderator Phoebe Yang, Board Director of GE Healthcare, Doximity, and CommonSpirit Health, and former General Manager, Amazon Web Services, and discussion leaders Susan Monarez, PhD, Deputy Director, Advanced Research Projects Agency for Health (APRA-H); Cris Ross, Chief Information Officer, Mayo Clinic; and Suchi Saria, PhD, Founder & CEO, Bayesian Health, explored how executives can best lead their organizations through this unprecedented time of innovation.
Don’t let fear get in the way of transformation
Leaders recognized AI’s potential to revolutionize health care practices and outcomes. Ross noted the impact language-based AI models could have in supporting providers’ relationships with patients and caregivers. “Medicine is about dialogue between a caregiver and a patient. If we can get to a language-based interface that allows providers to focus on their patients instead of having to click boxes in an electronic health record, it will change everything,” he said.
But while executives across the health care industry are excited about AI’s promise, they’re also concerned about the potential risks of integrating AI solutions into their practices. Ross shared that, in his role, he has “never felt that tension more acutely than right now, which is the fear of missing out, balanced with the fear of doing damage or litigation or regulatory breakage.”
Because of that tension, one of the challenges to advancing AI in health care is ensuring leaders have the information and expertise needed to move forward. “We need to think about training and channeling the excitement so that it doesn’t get overtaken by fear,” Yang said.
Saria agreed. “There’s a gap between expertise and research and best practices” in AI deployment “that exist in narrow pockets” but aren’t widely known, “and in the absence of that, we’re seeing leaders stop in their tracks,” she said. To address the issue, Saria said AI experts in the industry need to further disseminate resources and best practices in implementing AI in health care to empower executives and operational leaders to embrace AI solutions. Additionally, there needs to be reflection on and identification of the use-cases that already have been proven in studies, she said, adding that there are high impact use-cases where the potential is well understood that make for good starting points.
Saria also noted there are several initiatives that have published frameworks and guidelines for how to create and implement AI solutions for health care in ways that minimize risk. She highlighted efforts by the Coalition for Health AI, the National Academy of Medicine, and the National Institute of Standards and Technology as examples. “We now have pretty mature, good starting points for guardrails that organizations can look to, and we know of delivery organizations that are setting up governance committees and looking to these initiatives to understand how they can start thinking about using AI and incorporating AI solutions into their processes,” she said.
Start with governance and collaboration
Yang noted that health care leaders already have many of the mechanisms needed to take next steps. “Organizations have ethics groups and cybersecurity groups and other components needed to start evaluating and implementing AI solutions. They need to start to bring those together and apply them to AI, like they do to everything else, and be aware and stay informed,” she said.
However, Yang added, “no one can do it alone as an island. It takes collaboration and partnership.” She continued, “You have to have partnerships that fit into your ecosystem of trust and put guardrails up in a way that allows us to unleash all that AI has to offer.”
Saria said leaders should look “for partners that really have an understanding of implementing and using AI solutions from end-to-end.” These partnerships can help organizations “understand the governance, the valuation, the implementation, and adoption of the technology and enable your organization to build the muscle to continue moving forward,” she explained.
Lead with a realistic, cohesive, and ethical approach
Once organizations have governance structures and partnerships in place to advance their adoption of AI solutions, executives must be mindful to lead with a realistic, cohesive, and ethical approach to integrating them, the discussion leaders said.
To do so, Ross said executives must have practical expectations about the standards they set for AI solutions’ impact. “If our standard is that machines need to operate at a deterministic perfect level, when the rest of the ecosystem does not, we will have totally failed,” he said.
Yang agreed. “I love the cutting edge, but more than 90% of what we do in health care doesn’t need the latest-breaking cutting-edge AI models to be effective or make progress. What we need are reliable, tested AI-enabled solutions, and there are a lot of capabilities out there that we’re not leveraging that could make health care a lot more efficient, affordable, and effective,” she said.
Ross said leaders also must think strategically about how they can leverage different AI technologies to enhance their operations cohesively. Otherwise, organizations could run the risk of creating fragmentation and increasing burden on frontline providers.
“I’ve got doctors every day who have algorithms that they think can help avoid disease or diagnose disease in cheaper, faster, better ways, and they’re dying to get it to the bedside,” Ross said. “But we’re trying to figure out how to do that in a way that it’s safe, and also in a way that doesn’t make what is already an incredibly complex information space for our doctors harder,” he explained.
Focusing on “concrete use cases that are important across the board” is one way executives can ensure they’re leading with a strategic approach, Saria said. She suggested that leaders conduct “a risk-benefit analysis” to determine whether an AI solution will be “a net positive in their organization.” Those analyses can help executives to “break AI strategy down into the concrete areas and use cases where AI can work and be evaluated,” she said.
But as organizations begin to adopt new solutions, they must be mindful to address ethical issues that arise, the discussion leaders said.
Monarez stressed the need to broaden AI solutions’ reach to ensure that everyone benefits from advancements, rather than further exacerbating current disparities. She noted that leaders should be cognizant of scaling and leveraging AI solutions to help address inequities and make health care more accessible for everyone. The onus falls on health care leaders to bridge the gap and prevent the digital divide that could leave underserved populations behind, Monarez said.
That’s why ARPA-H launched its Biomedical Data Fabric Toolbox initiative, Monarez said. “When you have advanced tools like AI, you have to leverage those to create better health outcomes,” but “you cannot innovate if you don’t have data,” she noted. The Biomedical Data Fabric Toolbox initiative seeks to advance “opportunities to harvest the zettabytes of data that exist within federal health care programs and throughout the industry and make them integrated, interoperable, and useful for all who are trying to do positive disruptive innovation across the health care industry,” Monarez said.
With these considerations in mind, the discussion leaders were optimistic that health care executives can continue to move the industry forward by harnessing advancement in AI. “I am very hopeful that there are credible leaders who really can navigate and lead the way, and very concrete pragmatic partnerships to build on,” Saria said.