|
Signal of the week
At the HIMSS AI Forum, senior health IT leaders warned that AI adoption is creating two tiers of clinical care, and the gap is accelerating
At the HIMSS AI in Healthcare Forum in Boston on June 25, 2026, Dr. Michael Pfeffer, Chief Information and Digital Officer at Stanford Healthcare and Associate Dean at Stanford School of Medicine, described the disparity directly. An ambulance reaching a well-funded hospital gives its patient access to AI-enabled diagnostic support and real-time clinical decision tools that the same patient in the other ambulance, heading to a smaller, resource-constrained facility, would not receive. Pfeffer called it an unfair division in patient care.
Dr. Eric Alper, Chief Quality Officer and CCIO at UMass Memorial Health, named the structural cause: smaller, safety-net hospitals and independent clinics lack the capital, technical staff, and data infrastructure to build, experiment with, and monitor AI tools.
The data behind the concern:
According to ASTP/ONC analysis of the AHA IT Supplement survey, 71% of nonfederal acute care hospitals reported using predictive AI integrated into their EHRs in 2024, up from 66% in 2023. The authors describe a “persistent digital divide”: smaller, rural, independent, and critical access hospitals remain significantly behind larger, system-affiliated peers.
The validation gap amplifies the adoption gap: a PLOS Digital Health study found only 61% of hospitals that have deployed AI conduct local evaluations for accuracy, and only 44% assess tools for bias across patient subgroups. For smaller hospitals, with less internal technical capacity, these percentages are lower still.
|
The so what
If you lead a smaller or community health system: the practical near-term path to AI access runs through EHR vendors and health system consortia, not internal development. Ask your EHR vendor directly what AI tools they are deploying that do not require you to build your own evaluation capacity. If you lead a larger system: the AI divide is also your problem; your referral networks and regional care coordination depend on the capabilities of the facilities around you. A well-resourced academic medical center whose regional referring hospitals cannot afford to validate AI tools will find the quality gap showing up in the patients who arrive too late.
|
|
|
The clinical evidence check
Mixed — directional evidence exists; controlled studies on the gap's clinical impact are lacking
Does the AI adoption gap between health system types create measurable differences in patient outcomes?
Evidence that AI tools improve outcomes at the systems deploying them is growing. Studies of AI-assisted sepsis detection have found materially lower sepsis mortality at institutions with fully integrated AI early warning systems. AI-assisted radiology reads have shown lower miss rates for actionable findings at institutions with workflow-integrated systems.
What has not been studied directly: whether the growing gap in AI adoption between large and small health systems is producing measurable outcome differences at the patient population level. Controlled studies comparing outcomes at AI-adopter versus non-adopter hospitals of comparable size and patient complexity do not yet exist in sufficient volume to draw causal conclusions.
The equity evidence is more developed: a RAND Corporation analysis found only 17% of U.S. hospitals using AI reported performing any equity impact assessment before deployment. Smaller hospitals are overrepresented in the 83% that do not assess for equity impact; raising the risk that tools adopted without validation may perform worse for the patient populations that smaller hospitals disproportionately serve.
|
|
Regulatory radar
| CMS |
The Rural Health Transformation Program provides states with funding to expand virtual care access, modernize health infrastructure, and adopt digital tools in rural settings. Implementation timelines vary significantly by state. Health systems in rural states should verify their state's RHTP implementation status and any funding available for digital health tool adoption. |
| HHS |
ASTP/ONC's response to the HHS AI RFI explicitly identified the AI adoption divide as a reason for cross-agency coordination to support resource-scarce health systems. HHS committed to cross-agency AI adoption support, though specific programs targeting smaller health systems have not yet been announced. |
| FDA |
FDA's January 6, 2026 guidance reduced oversight for certain low-risk clinical decision support and digital health tools, allowing more AI tools to reach clinics without full premarket review. For smaller health systems, this means more tools will be available without FDA clearance, requiring those systems to exercise their own validation judgment rather than relying on FDA clearance as a quality signal. |
|
|
Practitioner voice
|
Dr. Eric Alper
CCIO & Chief Quality Officer, UMass Memorial Health · HIMSS AI Forum, June 25, 2026
|
“How are they going to be able to access these kinds of tools, which will transform healthcare? We're creating a digital divide for the haves and have-nots in the community.”
Alper's proposed path forward: EHR vendors need to play a larger role in delivering AI tools to health systems that cannot build them independently. “We do rely on our EHR vendors to help to deliver some of these tools to us.” For smaller health systems, the path to AI access runs through the vendor relationship, not the R&D budget.
|
