July 23, 2025 | Two of the great promises of AI is that it can help fill the gap in care in underserved communities and that it will give clinicians in high-demand specialties tools to meet that demand. One panel of experts at the HIMSS AI in Healthcare Forum in Brooklyn, NY, July 11-12, described how their organizations are implementing projects in exactly those areas.
The HIMSS AI in Healthcare Forum brought together IT leaders, clinicians, and administrators to discuss how AI is being implemented in healthcare systems across the country. The event was the first in a series called 2025 Future of AI in Healthcare, designed to help teams select technologies, build infrastructure, and create roadmaps for success with AI.
AI to Improve Care to the Underserved
The presentations, panels, and discussions included topics ranging from governance policies to frameworks for deciding whether to build custom AI models or purchase solutions. In the panel discussion titled “Safety Net Innovators: Bringing Digital Transformation to Underserved Hospitals,” representatives from the Health AI Partnership (HAIP) introduced the organization and representatives from the first member cohort pursuing AI projects using HAIP’s resources.
HAIP is a collaborative, seeking to empower healthcare organizations to use AI safely, effectively and ethically and provides resources, guidance, and standards. Their inaugural Practice Network includes five healthcare organizations:
- Health Center of Southeast Texas
- Community-University Health Care Center at the University of Minnesota
- North Country Healthcare
- San Ysidro Health
- WakeMed
Representatives from each organization participated in the panel to talk about the challenges and successes they are encountering in implementing individual AI projects. For example, at North Country Healthcare, one issue is that in many parts of their service area, internet service is weak, and another is that their EHR is outdated. At the Health Center of Southeast Texas, rapid growth creates issues. Both organizations are working to implement ambient AI scribes to take some of the pressure off overburdened providers.
"Artificial Intelligence, particularly through the integration of ambient AI technologies, is already enhancing operational workflows and reducing documentation burden" for some diagnosticians, says Jill Seys, DNP, digital health strategist, HIMSS. In facilities that serve un- and under-insured populations, lessening the burden on clinicians translates to better care for patients.
At San Ysidro Health, in San Diego County, the project is to use AI to screen for diabetic retinopathy. The system has a large population of medically underserved, low-income, and under-insured patients, along with high rates of diabetes mellitus (around 12%) and low rates (47%) of diabetic retinopathy screening. Diabetic retinopathy is asymptomatic, but early treatment is critical to preserve vision.
Their plan is to provide point-of-care AI to perform the screening so that it becomes accessible in the primary care setting, and is integrated with the system’s EHR. Integrating the screening into the primary care workflow with AI means that patients don’t need to visit a specialist, clinical decision-making is improved, and diabetic retinopathy is diagnosed earlier, improving patient outcomes.
This isn’t their first attempt to incorporate AI.
“We dug in [on AI] and failed. We were not ready. It wasn’t the vendor, it was us,” said Sonia Tucker, VP Population Health at San Ysidro. She said that this work with HAIP is the organization’s second attempt, and that between the first failed efforts and now, they had to make some changes.
She noted that the workplace culture needed some work, the CEO had to be convinced that AI is a real solution. Now, she says, “Our AI and data governance is in place. It’ s not perfect but we can implement more and make better decisions.”
Tucker’s comments followed themes that came up repeatedly throughout the Forum. Before implementing AI, healthcare organizations need to have appropriate governance policies, a receptive workforce, plans for on-going training, and a practical problem that needs to be solved.
For organizations that are part of the safety net, getting all of that groundwork in place can pose significant challenges. Funding, expertise, and other barriers can leave leaders in these organizations feeling like there’s an answer to some of their problems but it’s out of reach.
HAIP helps connect member organizations with resources, mentors, and standards. With thousands of healthcare clinics and facilities in the United States, most under-resourced, the need for AI is enormous, but without the right elements in place first, implementation will fail.
Along with the work HAIP is doing, other groups are coming together to tackle the issue around how to best deploy AI. “Subgroups and many different organizations that work in the health IT space are prompting questions about the right collaborations to ensure AI is implemented safely, with the right governance and organizational structures,” Sey notes.