Point-of-Care AI at SSM Health Boosts Diabetic Eye Screening 15-20% and 2.5x Among Lower-Income Patients

Point-of-Care AI Lifted Diabetic Retinopathy Screening at SSM Health

Diabetes is the second most common cause of irreversible vision loss in the U.S. One in four people with a new diabetes diagnosis already shows diabetic eye changes, yet 40-60% skip the recommended exam. National Eye Institute and CDC resources underscore the stakes: early detection prevents avoidable vision loss.

SSM Health, based in St. Louis, implemented autonomous AI for diabetic retinopathy screening and saw measurable gains. Screenings rose 15-20%, and individuals from lower socioeconomic groups became more than 2.5-times as likely to be screened.

The barrier: time, access, and competing priorities

"Our patients are scared of diabetes and even more scared of losing their vision," said SSM Health Medical Director Dr. Joseph Eckelkamp. "They see the value, but their circumstances get in the way."

Patients juggle multiple jobs, caregiving responsibilities, and transportation challenges. Many live in areas with few eye care professionals. When forced to choose between immediate needs and a preventive exam, they delay-often until symptoms surface.

The shift: bring screening to primary care

SSM Health implemented autonomous AI from Digital Diagnostics at the point of care. The goal: make screening a seamless part of routine diabetes visits and deliver results during the same encounter.

"Patients could receive their screening during a visit to their primary care provider-making it a true one-stop shop," Eckelkamp said. Immediate results gave the care team a clear path: schedule routine follow-up or refer to an eye specialist.

How SSM Health operationalized it

Topcon non-mydriatic fundus cameras were placed in primary care offices that also serve as key endocrinology referral hubs. Digital Diagnostics trained teams and provided software; clinics dedicated a dark room to improve diagnosability.

Workflows varied by team, but the core steps were consistent. Patients were flagged in pre-visit planning, and front desk or rooming staff completed the exam by standing order. Results flowed directly into Epic, so providers could act immediately.

To reinforce quality, the vendor reviewed images and shared feedback through the SphereDx platform. Teams used these insights to fix reversible errors and keep performance tight.

Results that moved the needle

• 15-20% increase in diabetic retinopathy screening across the system.
• 20% of screened patients had diabetic eye changes identified.
• People in lower socioeconomic groups were more than 2.5x as likely to be screened.

This approach also reduced friction for patients and staff. "We're ensuring optometrists and ophthalmologists are focused on treating disease rather than performing routine screenings," Eckelkamp said. "AI is helping deliver complex skill sets directly to the patient, enabling our teams to operate at the top of their license."

Why it worked

• One-visit convenience: screening happens during an established diabetes appointment.
• Near-instant results: no ambiguity about exam completion or next steps.
• EHR integration: results land in Epic for same-day planning.
• Clear roles: standing orders let non-physician staff run the exam.
• Continuous QA: image feedback loops sharpened team performance.

What you can copy in your system

• Equip primary care with non-mydriatic fundus cameras and an autonomous AI tool.
• Use standing orders so front desk or rooming staff can run scans without delays.
• Build a dark-room protocol to improve image success rates.
• Integrate with your EHR for immediate result delivery and referral workflows.
• Set up vendor-supported QA to reduce non-diagnostic images and retraining overhead.
• Track KPIs: completion rates, equity by socioeconomic group, positive findings, closed-loop referrals.

Impact on capacity and equity

Primary care teams captured patients who would have skipped the exam. Eye specialists shifted their time toward disease management instead of routine screening. The equity signal is clear: the model expanded access for those who typically face the highest friction.

"AI will automate time-consuming steps so we can better use our most valuable resources-our teams and our patients' time," Eckelkamp said. "The aim is simple: give people back the time that would otherwise be taken from them."

Bottom line for healthcare leaders

Point-of-care autonomous AI for diabetic retinopathy delivers measurable gains in screening rates, earlier detection, and equitable access. The model is operationally straightforward, friendly to primary care workflows, and aligned with top-of-license practice.

If you're building AI capability across clinical operations, explore practical upskilling paths: AI courses by job.