Going the Last Mile: Bringing Evidence to Bear on Healthcare AI Practice and Policy
At Digital Healthcare Research (DHR), the core of our work is research. The projects we fund close critical gaps in evidence about how digital healthcare technologies work in the real world. That evidence offers vital insights for healthcare delivery, clinical practice, and future studies, as well as for informing healthcare policy.
Across the healthcare enterprise today, the need for evidence-based insights is arguably most acute in the area of healthcare artificial intelligence (AI). From lawmakers and healthcare agencies to individual providers and patients, decision makers want to know how to seize the potential of healthcare AI while avoiding the pitfalls.
AI-enabled technologies are rapidly making their way into practice. For example, automations are streamlining scheduling and other administrative workflows, large language models (LLMs) are powering notetaking and transcription, and predictive analytics are informing clinical decisions. Known—and unknown—concerns about these technologies are significant, including risks to patient safety, health outcomes, and health equity.
Ongoing efforts to leverage AI in service of healthcare may be well intentioned, but they may not be evidence based. Comprehensive evidence takes time to generate and gather, and the development and deployment of healthcare AI present an ongoing challenge for the research community. This makes it even more important to maximize and accelerate the reach of the evidence that does exist and to raise awareness of gaps in that evidence.
At DHR, this is what we call the “last mile” of our work—making sure the research we produce is understood and used.
In the case of healthcare AI, it is particularly urgent that the evidence reaches people who bear the burden of crafting policies and guidelines for developers, providers, and others.
Amplifying the impact of research
In addition to scientific journals and our own publications (including this annual report), DHR seizes opportunities to speak directly with decision makers, including those at the highest levels of government. This year, our experts were invited to join federal partners and other thought leaders in several formative conversations about how our nation can harness the power of healthcare AI.
We briefed congressional members and staff on the Senate Committee on Health, Education, Labor, and Pensions, which is charged with defining a legislative framework for the future of healthcare AI. We shared insights with the President’s Cancer Panel on how AI and other technologies can support patient navigation. Our leaders serve on the Department of Health and Human Services AI Task Force, which was created under Executive Order 14110 to guide the responsible deployment and use of AI-enabled healthcare technologies. We also supported efforts to coordinate federal AI research under the Networking and Information Technology Research and Development Program.
Our participation extends beyond the federal government. As members of the Coalition for Health AI, we support complementary efforts to accelerate safe and equitable AI development and application. This year, our experts also supported a key meeting convened by the National Academy of Medicine’s Leadership Consortium to explore key questions about LLMs and generative AI in health and medicine.
Real-world studies, real-world insights
In joining these and other important conversations, DHR helps ensure that leaders in healthcare AI practice and policy are aware of and understand the available evidence.
Our perspectives are drawn from work with the principal investigators who develop and implement healthcare AI in real-world conditions. In examining the outcomes, they capture firsthand accounts of the barriers, challenges, and consequences—intended or unintended—that providers and patients experience.
Thus, we can offer pragmatic insights that balance the considerable hype around healthcare AI with the realities of implementation.
A responsive research agenda
These conversations also afford our leaders the opportunity to listen. The better we understand the specific concerns of policymakers and healthcare leaders around AI, the more relevant our research agenda can be.
Already we see several areas with enormous potential impact, including:
- Methods for evaluating LLMs and generative AI
- Patient-centered risk prediction
- AI literacy for clinicians, patients, and caregivers
Perhaps the most pressing question facing leaders in healthcare practice and policy is how AI solutions and systems already in the field are affecting patient safety.
In response to that need, in July 2024 we announced a new funding opportunity to understand the impact of AI deployments in healthcare delivery systems. These grants will enable researchers to explore breakthrough AI solutions and key factors for safe implementation, such as workflow integration and ongoing governance.
Evidence makes a difference
As AI innovation accelerates, novel healthcare applications emerge, and as more AI-based technologies reach the field, the need for targeted research is immense. Agencies and organizations leading the development of healthcare AI best practice and policy face complex decisions with numerous implications for patient safety, health outcomes, and health equity.
At DHR, we are committed to producing the evidence they need, and we stand ready to bring that evidence where it can make the most difference.
