What happened
The Institut national d'excellence en santé et en services sociaux (INESSS) reviewed AI for diabetic retinopathy screening for Québec's health ministry in February 2021. It found that the systems can pick out patients who need treatment by a specialist. It also found that newer systems perform about as well as human readers or below them, and that specificity varies more in real-world data. Two economic studies compared tele-screening with and without AI, in the UK and Singapore. INESSS judged that AI grading with EyeArt, Retmarker or SELENA+ seems efficient, because its cost savings would offset its lower specificity. In the UK study, replacing the first level of human grading saved £11.81 per correct result with Retmarker and £4.5 with EyeArt. The Singapore study projected avoided costs of 489 000 $, nearly 20% of the screening budget. In both studies, AI grading was cheaper and less effective than manual grading. INESSS cautions that the results may not transfer to Québec, which lacks an established screening program. It describes the versions in the UK study as expert systems and SELENA+ as a deep learning system.
Sources
-
Assessment report, by authors with no tie to the system.
Institut national d'excellence en santé et en services sociaux (INESSS). Télédépistage de la rétinopathie diabétique à l'aide d'une solution d'intelligence artificielle. État des connaissances. Québec, February 2021. Résumé, Efficience.
Deux études repérées évaluent l’IARD en comparaison avec le télédépistage sans l’IA (Royaume-Uni et Singapour). L’IARD (EyeArt, Retmarker, SELENA+) semble une technologie efficiente, puisque la réduction des coûts qu’elle engendre compenserait la diminution de l’efficacité (spécificité).
Two identified studies evaluate AIDR in comparison to tele-screening without AI (UK and Singapore). AIDR (EyeArt, Retmarker, SELENA+) seems to be an efficient technology since the cost reduction it generates would compensate for the decrease in efficacy (specificity).
Details
AI system
- Reported configuration
-
EyeArt 1.0 and Retmarker 0.8.2 in the UK study
- Workflow
-
AI replaced the first level of human reading in the modeled screening programs. Human graders read the images after the AI in the semi-automated model.
Money
- Kind of value
-
Projected costs avoided when semi-automated AI screening replaces manual grading
- Amount
-
Estimate: 489 000 $
- Recipients and payers
-
Singapore's health system
Follow-up
- Conditions for replication
-
The screening program is already well established, with care pathways that see referred patients on time. Sensitivity and specificity in local use match those reported in the studies.
Entry details
- Sector
-
Healthcare.
- Purpose
-
Eyenuk, Retmarker and EyRIS built EyeArt, Retmarker and SELENA+ to grade retinal photographs for diabetic retinopathy and flag patients who need a specialist.
Not yet checked: model, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, currency and prices, period and unit, realization, gross or net, included costs, comparison, method, return and payback, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). INESSS finds AI retinopathy screening cheaper per correct result (AIBD-0746). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0746.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)