Small Hospitals, Thin Margins, and a Bet on AI
With hundreds of rural hospitals financially vulnerable, a growing number are deploying AI in screening and documentation
In Waynesboro, Mississippi, Wayne General Hospital is the only hospital in the county. This month the hospital began using an AI-assisted cardiac detection system from Eko Health across its emergency department and primary care clinics. Clinicians record heart sounds with a digital stethoscope; the platform analyzes the recording in under a minute and flags signs of potential heart disease.
For hospitals like Wayne General, decisions about new technology are tied directly to whether the facility can continue serving its community.
The move comes as rural hospitals face sustained financial strain. The Chartis Group identified 417 rural hospitals as vulnerable to closure in its 2026 analysis, and more than 40% were operating at a loss. Median margins remain around 1-2%.
Since 2005, nearly 200 rural hospitals have closed nationwide, leaving many communities with limited access to inpatient and specialty care.
Workforce pressure compounds the challenge. The Association of American Medical Colleges projects a national physician shortfall that could reach about 86,000 by 2036. Primary care and several specialties are expected to be hit hardest, and rural counties tend to feel the shortages first.
AI is entering some rural hospitals in narrow roles tied directly to daily operations. A December 2025 survey in JAMA Network Open found 31.5% of U.S. hospitals reported using generative AI tools in 2024.
Adoption clustered in larger, system-affiliated and teaching hospitals, while independent and higher-Medicaid hospitals were less likely to report active plans. Rural hospitals often fall into those categories.
Stretching Limited Expertise
At Wayne General, the cardiac tool is added to routine exams. If the AI flags elevated risk, clinicians can order confirmatory testing or move the patient into a referral pathway sooner.
In counties with limited on-site specialty coverage, shortening that interval can affect how quickly patients receive advanced evaluation.
Sanford Health has built a virtual care center that links many small hospitals and clinics to remote specialists and uses technology-driven workflows to improve follow-up and access.
“For many of our patients, it’s virtual care or no care at all,” Chief Medical Officer of Virtual Care Dave Newman said in a Sanford release.
Small hospitals are making measured adjustments. CGH Medical Center completed a migration to Epic and activated built-in AI features to surface findings and assist with documentation.
Hospital leaders told industry outlets the system extracted discrete data from clinical notes and flagged incidental radiology findings. The hospital reported the platform identified 310 incidental findings in the first eight weeks and prompted follow-up.
The rollout also included ambient documentation tools designed to reduce after-hours charting.
Academic pilots show both potential and constraint. In Brewster, Washington, researchers and clinic staff ran a pilot using AI-generated text messages to boost lung-cancer screening.
Anna Zamora-Kapoor, who worked on the project, said the study demonstrated promise but also revealed operational limits.
“Artificial intelligence is a promising resource, but there’s limited knowledge about what happens when we try using AI in the real world,” she said.
“The technology is evolving, and it’ll get better, but we’re not there yet.”
Less than half of recipients opened the messages, highlighting practical outreach barriers.
These deployments focus on specific functions, such as screening, documentation, and follow-up, rather than broad experimental programs.
What the Budget Allows
Financial realities shape adoption decisions. With margins near zero and many hospitals operating at a loss, rural systems prioritize tools that reduce labor strain, support reimbursable care, or prevent avoidable transfers.
Policy alignment plays a role. The American Medical Association assigned a Category III CPT code for certain AI-assisted cardiac analyses (0962T). The Centers for Medicare & Medicaid Services later finalized national OPPS payment for Eko’s SENSORA platform.
Public disclosures report an assigned APC and a per-use payment of roughly $128.90 under that structure. That billing pathway makes point-of-care cardiac screening more financially feasible in outpatient settings.
There is no evidence yet that AI adoption has changed the broader trajectory of rural hospital closures. Most documented gains so far are operational, in faster note completion, improved follow-up, and earlier detection signals in defined use cases.
Those improvements can ease workflow pressure, but they do not address structural factors such as payer mix or long-term demographic shifts.
Infrastructure constraints persist. Broadband gaps, limited IT staffing, and capital limitations slow implementation. Many small hospitals adopt vendor-embedded features rather than building independent AI programs.
At Wayne General Hospital, the screening system adds information during the exam and informs next steps. In communities where specialty care may require hours of travel, earlier identification can influence referral timing, even if it does not resolve the deeper financial pressures rural hospitals face.