AI diabetic retinopathy screening in Northern Ontario: what the Thunder Bay research found

Researchers from Lakehead University, NOSM University, and Thunder Bay Regional Health Sciences Centre published peer-reviewed evidence in July 2026: AI-based diabetic retinopathy screening at a Thunder Bay primary care clinic cost 52% less per diagnosed case than the physician-led approach, with 93.6% of patients completing the AI exam without needing dilating drops. Here is what the study found and why it matters for eye care access in Northwestern Ontario.

A peer-reviewed study from Thunder Bay, published in July 2026 in BMJ Open Ophthalmology, provides the first Canadian cost evidence that autonomous AI can screen for diabetic retinopathy in a primary care setting at roughly half the cost of the traditional physician-based approach. The study ran at a Thunder Bay primary care clinic from December 2022 through October 2023. Among 202 adult patients with diabetes, 93.6% completed the AI exam without needing dilating drops, 22.2% were found to have referable retinopathy requiring ophthalmologist follow-up, and the cost per diagnosed case came to $620.36 Canadian — versus $1,283.55 for the conventional pathway, a 52% reduction. For Northwestern Ontario, where ophthalmologist access is limited by geography and physician shortages, the research is direct evidence that the tools to address the region's eye-care gap already exist in a locally tested form.

What the study measured

The study was led by Vishaal Bhambhwani of NOSM University and Thunder Bay Regional Health Sciences Centre, with co-authors Livio Di Matteo of Lakehead University and Noelle Whitestone of Orbis International. The AI used is an autonomous system that captures retinal images and identifies referrable diabetic retinopathy without requiring an ophthalmologist on site. At the Thunder Bay clinic, screening took place during a routine primary care appointment. The full cost-per-screening comparison across 100 patients was $C13,647.84 for the AI pathway versus $C28,238.09 for the physician pathway. The cost-per-diagnosed-case figure — $620.36 versus $1,283.55 — reflects the difference in detection yield between the two approaches on the same patient population. Patients flagged by the AI as having referable retinopathy were referred to an ophthalmologist for follow-up; the AI identified the cases requiring specialist attention, not the diagnosis or treatment itself. The study was published as a cost analysis; an earlier companion paper from the same research group, published in December 2024, reported on feasibility and patient experience at the same clinic.

The access gap this is designed to close

Diabetic retinopathy is a leading cause of preventable vision loss among working-age adults in Canada, and early detection through regular screening substantially reduces the risk of progression to serious sight loss. Screening is recommended annually for people with diabetes, but regular access depends on ophthalmologist availability — and in Northwestern Ontario, that availability is limited. The region has documented physician shortages across multiple specialties, and communities outside Thunder Bay often face long travel times to reach an ophthalmologist. Indigenous communities in the region carry a higher burden of diabetes than the national average, and geographic isolation compounds the screening gap. The existing complement — including mobile eye clinic programs such as the Vision Loss Rehabilitation Canada Eye Van, which travels over 6,000 kilometres annually in Northern Ontario — does critical work in this gap, but covers a vast geography. Autonomous AI screening at primary care clinics shifts the screening to the existing appointment, reaching patients who would not otherwise make a separate specialist trip. Health Canada's National Strategy for Eye Care explicitly cites AI pilot programs in regions of Canada such as northern Ontario as promising innovations for reaching underserved populations.

AI-NORTH and the broader regional research context

The Thunder Bay study is not an isolated project. The AI-NORTH Collaboratory, established in 2023 through a formal partnership between NOSM University, Lakehead University, the Temerty Centre for AI Research and Education in Medicine at the University of Toronto, and AMS, exists specifically to examine AI's role in Northern Ontario healthcare — across practice, policy, public health, education, and equity. Its mandate includes building a systematic picture of what AI tools are being studied and funded in northern and rural health contexts in Canada. The Thunder Bay diabetic retinopathy study is one of the peer-reviewed outputs that AI-NORTH's work is designed to produce and build on. The research base for AI in Northern Ontario healthcare is still early, and AI-NORTH represents the institutional infrastructure to develop it in a way that reflects Northern Ontario's specific geography, demographics, and health priorities.

What the evidence means in practice

For primary care clinics in Thunder Bay and surrounding communities, the most direct implication is that diabetic retinopathy screening can happen at the existing appointment, at a lower cost per detected case, using a system that requires no specialist presence. For health system administrators responsible for eye care access in the region, the Northern Ontario evidence base now exists in a peer-reviewed form that supports program planning conversations with the Ontario Ministry of Health and relevant health authorities. For patients with diabetes — particularly those in remote communities where a specialist referral currently means a significant trip — the practical change is a screening that does not require leaving home. None of that happens automatically from a study: the transition from published evidence to deployed program requires procurement, clinical workflow decisions, and policy commitment. The technical question, though, is now answered locally. A research team here ran the numbers in a Northern Ontario primary care setting, and the result is in the peer-reviewed record.

Sources: Screening for diabetic retinopathy with artificial intelligence in a primary care setting: a comparative cost analysis — Vishaal Bhambhwani (NOSM University / Thunder Bay Regional Health Sciences Centre), Livio Di Matteo (Lakehead University), Noelle Whitestone (Orbis International); BMJ Open Ophthalmology, Vol. 11, Issue 3, article e002904, July 30, 2026: pmc.ncbi.nlm.nih.gov/articles/PMC13422802/ | AI-NORTH Collaboratory — NOSM University, Lakehead University, Temerty Centre for AI Research and Education in Medicine (University of Toronto), AMS; mandate: inclusive and ethical AI innovation in Northern Ontario healthcare: arcandcentre.ca/research/merlin-research-groups/ai-north | National Strategy for Eye Care — Health Canada, citing AI pilot programs in regions including northern Ontario as promising innovations for reaching underserved populations: canada.ca/en/health-canada/services/publications/health-system-services/national-strategy-eye-care.html | Vision Loss Rehabilitation Canada Eye Van: visionlossrehab.ca — contextual reference only on mobile eye care in Northern Ontario