Screening for diabetic retinopathy with artificial intelligence in a primary care setting: a comparative cost analysis
Open accessAbstract
Objective: Cost analysis of autonomous artificial intelligence (AI)-based screening of diabetic retinopathy (DR) for adults with diabetes at a primary care clinic. Methods and analysis: This study provides a comparative cost analysis of actual results using AI-based DR screening with counterfactual results based on all patients going through the physician-based referral system. A cost analysis is conducted using cost data from published sources, provincial billing codes, statistical sources, and patient characteristics from a clinical study to compare autonomous AI-based screening for DR versus physician-based screening. Costs considered include direct costs of operating the AI system, physician fees, and indirect costs to patient time. Along with total cost comparisons, a cost per DR case detected is estimated and a sensitivity analysis based on variations in AI costs is provided. Results: Over the study period, 202 participants were screened for DR using autonomous AI. The majority (93.6%, n=189) of AI-based DR screening exams were completed successfully. The AI-based scenario results in total direct costs of $C7919.04 and indirect costs of $C5728.80, resulting in total costs of $C13 647.84 per 100 patients. The traditional physician-based approach results in total direct costs of $C8240 and indirect costs of $C19 998.09, resulting in total costs of $C28 238.09 for 100 patients. When costs are converted to costs per unit outcome, the total cost per diagnosed DR case is $C620.36 for the AI-based approach and $C1283.55 for the physician-based approach; the AI-based cost per diagnosed case was 52% lower. Conclusion: Given the lower cost per diagnosed case of the AI-based approach, there are advantages to the implementation of AI-based screening for DR.
AI snapshot
Written from the paper’s full text.
TL;DR
This comparative cost analysis used real-world autonomous AI-based diabetic retinopathy screening data from 202 primary care patients in Thunder Bay, Ontario, alongside a counterfactual physician-based referral scenario. It found lower total costs and lower cost per diagnosed DR case for AI-based screening than for traditional physician-based screening.
Key findings
- Among 202 participants, 93.6% (n=189) of AI-based DR screening exams were completed successfully; 22.2% (42/189) of completed exams detected referrable DR, and 55 patients were referred to ophthalmology, including 13 whose exams could not be completed.
- In the base case per 100 patients, AI-based screening had total costs of $C13,647.84 (direct $C7,919.04; indirect $C5,728.80), versus $C28,238.09 for traditional physician-based screening (direct $C8,240; indirect $C19,998.09).
- Cost per diagnosed DR case was $C620.36 for AI-based screening versus $C1,283.55 for physician-based screening, 52% lower. With OCT added, costs were $C634.58 versus $C1,336.05, 53% lower; with the AI fee 50% lower, AI was $C505.50 versus $C1,336.58, 62% lower.
- The paper reports that the bulk of the cost difference was in indirect costs to patients and caregivers, mainly income foregone and opportunity cost of time.
Methods
- Comparative cost analysis from health system and societal perspectives, comparing actual AI-based DR screening results with counterfactual results in which all patients went through the traditional physician-based referral system; the study states it generally conforms to CHEERS 2022 reporting guidelines.
- Data came from a prospective cohort of adults aged 18 years and older with type 1 or type 2 diabetes presenting for routine primary care in Thunder Bay, Ontario, from December 2022 through October 2023; 202 participants were screened using the iCare DRSPlus fundus imaging system and EyeArt autonomous AI.
- Costs included direct AI operating costs, physician fees, and indirect patient/caregiver time; results were estimated per 100 patients, with sensitivity analyses varying AI fees and adding retinal OCT.
Limitations
- The paper states that the study is based on a single primary care office, has a relatively small sample, and included predominantly older Caucasian patients, limiting generalisability to broader Canadian populations, remote communities, Indigenous populations, or younger working-age cohorts.
- The paper states that the cost analysis relies on assumptions about sunk costs, time foregone for appointments, and US billing codes for AI fees, with sensitivity analyses provided; it also notes that cost-effectiveness depends on AI test accuracy and population characteristics, and that the study does not evaluate diagnostic accuracy.
- Evident from the text: the traditional physician-based comparator is counterfactual rather than observed, and the AI fee was estimated from a US Medicare code because no Canadian reimbursement rate existed, introducing uncertainty.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
10 of 14 cited sources checked · 7 supported · 0 flagged · accuracy not yet measured
- Premisecitation
Using autonomous AI for diabetic retinopathy screening has been shown to be lower cost in other countries.
“using autonomous AI for DR screening has been shown to be lower cost in other countries.13–16”
Quote in the paperNot a substantial claim3 of 4 sources checked- cites [13] Xie, 2020, Artificial intelligence for teleophthalmology-based diabetic retinopathy screening in a national programme: an economic analysis modelling study, Lancet Digit Health source
- cites [14] Huang, 2022, Cost-effectiveness of artificial intelligence screening for diabetic retinopathy in rural China, BMC Health Serv Res source
- cites [15] Wolf, 2020, Cost-effectiveness of autonomous point-of-care diabetic retinopathy screening for pediatric patients with diabetes, JAMA Ophthalmol source
The Results report that autonomous AI screening was cost-saving compared with eye care professional screening when adherence exceeded 23%, but not in the 20% baseline scenario. The Discussion places the pediatric patient-and-caregiver model in a US context; it does not establish lower costs in other countries. (LLM, conf=0.88)
Experimental — accuracy not yet measured - cites [16] Hu, 2024, Population impact and cost-effectiveness of artificial intelligence-based diabetic retinopathy screening in people living with diabetes in Australia: a cost effectiveness analysis, EClinicalMedicine source
- Premisecitation
The AI-based approach has high levels of patient satisfaction.
“Given the cost per case advantage of the AI-based approach, combined with its feasibility and high levels of patient satisfaction, this suggests significant advantages to the implementation of AI-based screening for DR for both physicians and patients.”
Quote in the paperSubstantial claim1 of 1 sources checked- cites [6] Bhambhwani, 2025, Feasibility and Patient Experience of a Pilot Artificial Intelligence-Based Diabetic Retinopathy Screening Program in Northern Ontario, Ophthalmic Epidemiol source
- Premisecitation
Decreasing the number of patients requiring specialist time has been shown to increase specialist productivity.
“and has been shown to increase specialist productivity,30”
Quote in the paperNot a substantial claim1 of 1 sources checked- cites [30] Abramoff, 2023, Autonomous artificial intelligence increases real-world specialist clinic productivity in a cluster-randomized trial, NPJ Digit Med source
- Premisecitation
The AI-based approach is feasible.
“Given the cost per case advantage of the AI-based approach, combined with its feasibility and high levels of patient satisfaction, this suggests significant advantages to the implementation of AI-based screening for DR for both physicians and patients.”
Quote in the paperSubstantial claim1 of 1 sources checked- cites [6] Bhambhwani, 2025, Feasibility and Patient Experience of a Pilot Artificial Intelligence-Based Diabetic Retinopathy Screening Program in Northern Ontario, Ophthalmic Epidemiol source
- Findingown data
In the traditional physician-based approach, total direct costs were $C8240 per 100 patients.
“The traditional physician-based approach results in total direct costs of $C8240 (all for ophthalmologist fees) and indirect costs of $C19 998.09, resulting in total costs of $C28 238.09 for 100 patients.”
Quote in the paperSubstantial claim- The authors state this as an assumption: Sunk costs for both the AI-based and physician-based approaches are zero.
- Findingown data
In the AI-based scenario, indirect costs were $C5728.80 per 100 patients.
“The AI-based scenario results in total direct costs of $C7919.04 ($C5686 for AI screening and $C2233 for ophthalmologist fees) and indirect costs of $C5728.80, resulting in total costs of $C13 647.84 per 100 patients.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The remaining 5.5% of participants took public transit.
- Findingown data
In the traditional physician-based approach, total costs (direct and indirect) were $C28,238.09 per 100 patients.
“The traditional physician-based approach results in total direct costs of $C8240 (all for ophthalmologist fees) and indirect costs of $C19 998.09, resulting in total costs of $C28 238.09 for 100 patients.”
Quote in the paperSubstantial claim - Findingown data
In the traditional physician-based approach, indirect costs were $C19,998.09 per 100 patients.
“The traditional physician-based approach results in total direct costs of $C8240 (all for ophthalmologist fees) and indirect costs of $C19 998.09, resulting in total costs of $C28 238.09 for 100 patients.”
Quote in the paperSubstantial claim - Findingreasoning
In the AI-based scenario, total costs (direct and indirect) were $C13,647.84 per 100 patients.
“The AI-based scenario results in total direct costs of $C7919.04 ($C5686 for AI screening and $C2233 for ophthalmologist fees) and indirect costs of $C5728.80, resulting in total costs of $C13 647.84 per 100 patients.”
Quote in the paperSubstantial claim- The authors state this as an assumption: Sunk costs for both the AI-based and physician-based approaches are zero.
- The authors state this as an assumption: The remaining 5.5% of participants took public transit.
- Findingreasoning
In the modified base case including OCT, the total cost per diagnosed diabetic retinopathy case was $C1336.05 for the traditional physician-based approach.
“In this case, the total cost per diagnosed DR case is $C634.58 for the AI-based approach and $C1336.05 for the traditional physician-based approach.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
- Findingreasoning
In the revised AI fee scenario, the total cost per diagnosed diabetic retinopathy case via the AI-based approach was 62% lower than the traditional physician-based approach.
“In this final scenario, the total cost per diagnosed case of DR via the AI-based approach is now 62% lower than the traditional physician-based approach.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
- Findingown data
Among the 202 participants screened, 55 were referred to an ophthalmologist for further review.
“Among the 202 participants screened, 55 were referred to the ophthalmologist for further review.”
Quote in the paperNot a substantial claim - Findingown data
The AI-based cost per diagnosed diabetic retinopathy case was 52% lower than the traditional physician-based approach in the base case.
“The AI-based cost per diagnosed case was 52% lower.”
- Also stated in: § Discussion · plain
“The results of this comparative analysis show that the AI-based approach has a lower cost per case than the traditional physician-based approach when screening for DR.”
Quote in the paperSubstantial claim - Also stated in: § Discussion · plain
- Findingreasoning
In the modified base case including OCT, the AI-based cost per diagnosed diabetic retinopathy case was 53% lower than the traditional physician-based approach.
“The AI-based cost per diagnosed case was 53% lower.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
- Findingown data
93.6% (189/202) of AI-based diabetic retinopathy screening exams were completed successfully.
“The majority (93.6%, n=189) of AI-based DR screening exams were completed successfully with only 3 participants (1.56%) requiring mydriatic eyedrops for imaging.”
Quote in the paperNot a substantial claim - Findingreasoning
In the modified base case including OCT, the total cost per diagnosed diabetic retinopathy case was $C634.58 for the AI-based approach.
“In this case, the total cost per diagnosed DR case is $C634.58 for the AI-based approach and $C1336.05 for the traditional physician-based approach.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
- Findingown data
The total cost per diagnosed diabetic retinopathy case was $C620.36 for the AI-based approach.
“When costs are converted to costs per case outcome, the total cost per diagnosed DR case is $C620.36 for the AI-based approach and $C1283.55 for the traditional physician-based approach.”
Quote in the paperSubstantial claim - Findingreasoning
The total cost per diagnosed diabetic retinopathy case was $C1283.55 for the traditional physician-based approach.
“When costs are converted to costs per case outcome, the total cost per diagnosed DR case is $C620.36 for the AI-based approach and $C1283.55 for the traditional physician-based approach.”
Quote in the paperSubstantial claim - Findingown data
Over the study period, 202 participants were screened for diabetic retinopathy using autonomous artificial intelligence.
“Over the study period, 202 participants were screened for DR using autonomous AI.”
Quote in the paperSubstantial claim - Findingown data
In the revised AI fee scenario, the total cost per diagnosed diabetic retinopathy case for the AI-based approach fell from $C620.36 to $C505.50, a 19% reduction.
“As a result, the total cost per diagnosed case of DR falls from $C620.36 to $C505.50 for the AI-based scenario—a 19% reduction.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
- Findingown data
In the AI-based scenario, total direct costs were $C7919.04 per 100 patients.
“The AI-based scenario results in total direct costs of $C7919.04 ($C5686 for AI screening and $C2233 for ophthalmologist fees) and indirect costs of $C5728.80, resulting in total costs of $C13 647.84 per 100 patients.”
Quote in the paperSubstantial claim- The authors state this as an assumption: Sunk costs for both the AI-based and physician-based approaches are zero.
- Findingown data
Additional sensitivity results calculated assuming lower hours per visit and lower hours of work and income foregone combined with higher AI fees support the AI-based approach having a lower cost per case relative to the physician-based approach.
“As well, additional results were calculated assuming lower hours per visit and lower hours of work and income foregone combined with higher AI fees, and these results also support the AI-based approach having a lower cost per case relative to the physician-based approach.”
Quote in the paperNot a substantial claim - Findingown data
Costs per diagnosed case of diabetic retinopathy ranged from 52% to 62% lower for the AI-based approach relative to the physician-based approach.
“Costs per diagnosed case of DR ranged from 52% to 62% lower for the AI-based approach relative to the physician-based approach.”
Quote in the paperSubstantial claim - Interpretationreasoning
The bulk of the cost difference between the AI-based and traditional approaches in the base case is in indirect costs to patients and caregivers (income foregone and opportunity cost of time).
“When reviewing the base case, the bulk of the cost difference is in indirect costs to patients and caregivers in terms of income foregone and opportunity cost of their time, given that in the AI approach a much smaller proportion require an additional visit for specialist screening.”
Quote in the paperSubstantial claim - Interpretationreasoning
The differences in cost per case between the AI-based and traditional approaches are more rooted in costs to patients than in direct healthcare expenditures.
“Thus, it should be acknowledged that the differences in cost per case are more rooted in costs to patients rather than direct healthcare expenditures.”
Quote in the paperSubstantial claim - Interpretationreasoning
The difference in cost per diagnosed case is mainly due to the large difference in indirect time-based costs to patients.
“Again, the difference is mainly due to the large difference in indirect time-based costs to patients—which is something that is of importance to patients based on their survey results ranking their experience with the AI approach in the original published study.”
Quote in the paperSubstantial claim - Study design and conductown data
The study is based on data from a single primary care office.
“First, it is based on data from a single primary care office.”
Quote in the paperSubstantial claim - Interpretationown data
In the revised AI fee scenario, the AI-based approach represents substantial savings in terms of both direct and indirect costs.
“In this scenario, the AI-based approach represents substantial savings in terms of both direct as well as indirect costs.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
- Interpretationcitation
Given projected shortages of ophthalmologists in Canada and long wait lists, decreasing the number of patients requiring specialist time relieves additional health system challenges.
“Given the projected shortages of ophthalmologists28 in Canada and long wait lists,29 decreasing the number of patients requiring specialist time relieves additional health system challenges”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [28] Bellan, 2016, Future trends in ophthalmology health human resources in Canada, Can J Ophthalmol source
- cites [29] Moir M , Barua B . Waiting your turn: wait times for health care in canada, 2023 report [Fraser Institute]. Available: https://www.fraserinstitute.org/sites/default/files/waiting-your-turn-2023-execsum.pdf [Accessed 20 Aug 2024].
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Interpretationreasoning
Generalisability of the study results to broader Canadian populations, remote communities, Indigenous populations or younger working-age cohorts is limited.
“Generalisability to broader Canadian populations, remote communities, Indigenous populations or younger working-age cohorts is therefore limited.”
Quote in the paperNot a substantial claim
- May combine two claims. 7
- Classified as probably not a substantial claim. 9
May combine two claims.
PremisecitationAn initial study in Australia demonstrated net savings for the health system and the potential for improved equity in access to care and health outcomes from AI-based diabetic retinopathy screening.
“Notably, one of the initial studies to evaluate cost-effectiveness of AI-based DR screening in Australia, a country with universal, publicly funded healthcare similar to Canada, demonstrated net savings for the health system and the potential for improved equity in access to care and health outcomes.16”
Quote in the paperNot a substantial claim1 of 1 sources checked- cites [16] Hu, 2024, Population impact and cost-effectiveness of artificial intelligence-based diabetic retinopathy screening in people living with diabetes in Australia: a cost effectiveness analysis, EClinicalMedicine source
May combine two claims.
PremisecitationPatients like not having to wait or come in for an additional specialist visit and not having to use dilating eyedrops with its consequences on their productivity for the next few hours.
“Patients like not having to wait or come in for an additional specialist visit and not having to use dilating eyedrops with its consequences on their productivity for the next few hours.6”
Quote in the paperSubstantial claim1 of 1 sources checked- cites [6] Bhambhwani, 2025, Feasibility and Patient Experience of a Pilot Artificial Intelligence-Based Diabetic Retinopathy Screening Program in Northern Ontario, Ophthalmic Epidemiol source
The Results report that participants highly valued not waiting for a separate eye-exam visit and not using dilating drops; the Discussion links both to satisfaction. The paper also says some participants refused drops because they needed to drive home, but it does not report an effect on productivity over the next few hours. No page numbers are available in our copy. (LLM, conf=0.87)
Experimental — accuracy not yet measured
May combine two claims.
PremisecitationThe autonomous AI screening strategy has high sensitivity and good specificity for the detection of referrable diabetic retinopathy compared with the traditional ophthalmologist approach.
“Our study, the first based in Canada, demonstrates a screening strategy using autonomous AI that has a high sensitivity and good specificity for the detection of referrable DR,27 compared with the traditional ophthalmologist approach, and is shown to result in lower cost per case diagnosis.”
Quote in the paperSubstantial claim1 of 1 sources checked- cites [27] Lim, 2023, Artificial Intelligence Detection of Diabetic Retinopathy: Subgroup Comparison of the EyeArt System with Ophthalmologists’ Dilated Examinations, Ophthalmol Sci source
Probably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductcitationThe autonomous AI platform used (EyeArt by Eyenuk) was Health Canada-approved.
“Images were uploaded to a Health Canada-approved autonomous AI platform, EyeArt by Eyenuk. (Woodland Hills, California, USA),18”
Quote in the paperNot a substantial claim1 of 1 sources checked- cites [18] Ipp, 2021, Pivotal Evaluation of an Artificial Intelligence System for Autonomous Detection of Referrable and Vision-Threatening Diabetic Retinopathy, JAMA Netw Open source
The Introduction identifies EyeArt as an Eyenuk system, and the Methods describe it as an autonomous AI system. The Introduction says EyeArt is FDA-cleared, but our copy does not state that it was approved by Health Canada; the approval claim is unsupported. (LLM, conf=0.88)
Experimental — accuracy not yet measured
May combine two claims.
Findingown dataIn the revised AI fee scenario (50% lower AI fee plus OCT), the AI-based direct costs were $C5392.21, a 32% reduction from $C7919.04.
“The results as presented in online supplemental appendix A2 show that the direct costs of the AI-based approach are now $C5392.21—a 32% reduction from $C7919.04.”
Quote in the paperSubstantial claim- Rests on an assumption: The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
- The authors state this as an assumption: Up to one-third of patients need to be scanned with OCT.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationDiabetic retinopathy screening programmes have been shown to be cost-effective.
“DR screening programmes have been shown to be cost-effective,26”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [26] source
May combine two claims.
Study design and conductown dataThe sample size is relatively small and the cohort included predominantly older Caucasian patients.
“The sample size is relatively small and our cohort included predominantly older Caucasian patients.”
Quote in the paperNot a substantial claimMay combine two claims.
Interpretationown dataThis study is the first based in Canada to demonstrate a screening strategy using autonomous AI that results in lower cost per case diagnosis.
“Our study, the first based in Canada, demonstrates a screening strategy using autonomous AI that has a high sensitivity and good specificity for the detection of referrable DR,27 compared with the traditional ophthalmologist approach, and is shown to result in lower cost per case diagnosis.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).
Interpretationnone givenAI-based screening has the potential to improve access to care for more remote and under-served communities.
“AI-based screening also has the potential to improve access to care for more remote and under-served communities.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Study design and conductnone givenThe study generally conforms to the Consolidated Health Economic Evaluation Reporting Standards 2022 reporting guidelines for economic evaluations.
“This study generally conforms to the Consolidated Health Economic Evaluation Reporting Standards 2022 reporting guidelines for economic evaluations.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Interpretationnone givenThe cost-effectiveness of the AI-based approach is dependent both on the sensitivity, specificity, positive and negative predictive values of the AI platform and on the characteristics of the population being screened.
“Third, the cost-effectiveness of the AI-based approach is dependent both on the sensitivity/specificity/positive and negative predictive values of the AI platform as well as the characteristics of the population being screened.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
InterpretationreasoningScreening a younger population with a higher employment rate would likely mean additional savings in terms of both direct and indirect costs.
“While the mean age of the population screened in this study was 70.5 years, screening a younger population with a higher employment rate would likely mean additional savings in terms of both direct and indirect costs.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S4 (based).
Interpretationnone givenThis study is one of the first comparative cost studies on AI-based diabetic retinopathy detection in Canada using real-world data.
“Despite these limitations, our study is still one of the first comparative cost studies on AI-based DR detection in Canada using real-world data, which can support the increased uptake and implementation of AI-based DR screening programmes in the country.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Study design and conductnone givenPatients aged 18 years and older with type 1 or type 2 diabetes presenting to a primary care provider in Thunder Bay, Ontario, from December 2022 through October 2023 were consecutively offered a screening exam for diabetic retinopathy using autonomous AI.
“Patients aged 18 years and older, diagnosed with type 1 or type 2 diabetes, presenting to their primary care provider for a routine appointment at a clinic in Thunder Bay, Ontario, from December 2022 through October 2023 were consecutively offered a screening exam for DR using autonomous AI.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductown dataThe study does not evaluate the diagnostic accuracy of the AI programme used.
“Our study does not evaluate the diagnostic accuracy of the AI programme used; however, as discussed in the manuscript, these have been widely published in the literature before.”
Quote in the paperNot a substantial claim
The authors state this as an assumption.
Sunk costs for both the AI-based and physician-based approaches are zero.
“Thus realistically, both physician-based and AI approaches have sunk costs; we are assuming they are zero for both.”
The paper relies on this without support.
The AI system detects macular oedema without any additional costs, whereas the traditional approach incurs an additional OCT cost.
“The AI system detects macular oedema without any additional costs whereas for the traditional approach, an additional cost for the OCT is incurred.”
The authors state this as an assumption.
Up to one-third of patients need to be scanned with OCT.
“This is not needed for all patients, but conservatively we can assume that up to one-third of patients need to be scanned.24”
The authors state this as an assumption.
The remaining 5.5% of participants took public transit.
“of which 94.5% (n=189) arrived using a private vehicle (and we will assume the remaining 5.5% took public transit).”
The paper relies on this without support.
Decreasing the number of patients requiring specialist time increases adherence to recommended screening guidelines.
“while also increasing adherence to recommended screening guidelines.”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Two different mean ages are reported for the cohort
MinorThe results and Table 1 report a mean age of 70.8 years, but the limitations section says it was 70.5 years.
“mean age 70.8 years (SD±11.7)”
“the mean age of the population screened in this study was 70.5 years”
Indigenous percentage does not match the count
MinorOf 202 participants, 5 is 2.48%, not 2.98%.
Reported 2.98 · recomputed 2.475248
“| Indigenous | 5 (2.98%) |”
Percentage requiring eyedrops does not match the count
MinorThree of 202 participants is approximately 1.49%, rather than 1.56%.
Reported 1.56 · recomputed 1.485149
“only 3 participants (1.56%) requiring mydriatic eyedrops for imaging.”
Traditional-approach costs change when only the AI fee is revised
MinorThe revised calculation is said to change only the AI test fee relative to the modified base case. Nevertheless, Table 3 changes the traditional approach's direct costs from $C9395.00 to $C9406.67.
“all the assumptions are the same as the modified base case except that the AI-based test fee is assumed to be 50% lower at $C28.43.”
“| Direct costs | 8240.00 | 9395.00 | 9406.67 |”
Cites
Global prevalence and major risk factors of diabetic retinopathy
Yau, J. W. Y., Rogers, S. L., Kawasaki, R., Lamoureux, E. L., Kowalski, J. W., Bek, T., Chen, S.-J., Dekker, J. M., Fletcher, A., Grauslund, J., et al. (2012) · Diabetes Care · doi:10.2337/dc11-1909 · Cited by 4,985 · see citing works
Early Photocoagulation for Diabetic Retinopathy
* (1991) · Ophthalmology · doi:10.1016/s0161-6420(13)38011-7 · Cited by 1,540 · see citing works
Photocoagulation Treatment of Proliferative Diabetic Retinopathy
The Diabetic Retinopathy Study Research Group (1981) · Ophthalmology · doi:10.1016/s0161-6420(81)34978-1 · Cited by 981 · see citing works
Cost-Effectiveness of Detecting and Treating Diabetic Retinopathy
Javitt, J. C., & Aiello, L. P. (1996) · Annals of Internal Medicine · doi:10.7326/0003-4819-124-1_part_2-199601011-00017 · Cited by 411 · see citing works
Xie, Y., Nguyen, Q. D., Hamzah, H., Lim, G., Bellemo, V., Gunasekeran, D. V., Yip, M. Y. T., Qi Lee, X., Hsu, W., Li Lee, M., et al. (2020) · The Lancet Digital Health · doi:10.1016/s2589-7500(20)30060-1 · Cited by 288 · see citing works
Ipp, E., Liljenquist, D., Bode, B., Shah, V. N., Silverstein, S., Regillo, C. D., Lim, J. I., Sadda, S., Domalpally, A., Gray, G., et al. (2021) · JAMA Network Open · doi:10.1001/jamanetworkopen.2021.34254 · Cited by 256 · cited 1× · see citing works
“(Woodland Hills, California, USA), 18 which produced a report recommending referral to an ophthalmologist when more than mild retinopathy was detected.”
Keel, S., Lee, P. Y., Scheetz, J., Li, Z., Kotowicz, M. A., MacIsaac, R. J., & He, M. (2018) · Scientific Reports · doi:10.1038/s41598-018-22612-2 · Cited by 220 · see citing works
Artificial Intelligence Detection of Diabetic Retinopathy
Lim, J. I., Regillo, C. D., Sadda, S. R., Ipp, E., Bhaskaranand, M., Ramachandra, C., Solanki, K., Dubiner, H., Levy-Clarke, G., Pesavento, R., et al. (2023) · Ophthalmology Science · doi:10.1016/j.xops.2022.100228 · Cited by 157 · cited 1× · see citing works
“16 Our study, the first based in Canada, demonstrates a screening strategy using autonomous AI that has a high sensitivity and good specificity for the detection of referrable DR, 27 compared with the traditional ophthal-mologist approach, and is shown to result in lower cost per case diagnosis.”
Cost-effectiveness of artificial intelligence screening for diabetic retinopathy in rural China
Huang, X.-M., Yang, B.-F., Zheng, W.-L., Liu, Q., Xiao, F., Ouyang, P.-W., Li, M.-J., Li, X.-Y., Meng, J., Zhang, T.-T., et al. (2022) · BMC Health Services Research · doi:10.1186/s12913-022-07655-6 · Cited by 89 · see citing works
Scheetz, J., Koca, D., McGuinness, M., Holloway, E., Tan, Z., Zhu, Z., O’Day, R., Sandhu, S., MacIsaac, R. J., Gilfillan, C., et al. (2021) · Scientific Reports · doi:10.1038/s41598-021-94178-5 · Cited by 88 · see citing works
Wolf, R. M., Channa, R., Abramoff, M. D., & Lehmann, H. P. (2020) · JAMA Ophthalmology · doi:10.1001/jamaophthalmol.2020.3190 · Cited by 87 · see citing works
Abramoff, M. D., Whitestone, N., Patnaik, J. L., Rich, E., Ahmed, M., Husain, L., Hassan, M. Y., Tanjil, M. S. H., Weitzman, D., Dai, T., et al. (2023) · npj Digital Medicine · doi:10.1038/s41746-023-00931-7 · Cited by 81 · see citing works
Self-citationWang, Y., Liu, C., Hu, W., Luo, L., Shi, D., Zhang, J., Yin, Q., Zhang, L., Han, X., & He, M. (2024) · npj Digital Medicine · doi:10.1038/s41746-024-01032-9 · Cited by 76 · see citing works
Fuller, S. D., Hu, J. Q., Liu, J. C., Gibson, E., Gregory, M. H., Kuo, J., & Rajagopal, R. (2020) · Journal of Diabetes Science and Technology · doi:10.1177/1932296820967011 · Cited by 60 · see citing works
Hu, W., Joseph, S., Li, R., Woods, E., Sun, J., Shen, M., Jan, C. L., Zhu, Z., He, M., & Zhang, L. (2024) · eClinicalMedicine · doi:10.1016/j.eclinm.2023.102387 · Cited by 47 · cited 2× · see citing works
“16 Notably, one of the initial studies to evaluate cost-effectiveness of AI-based DR screening in Australia, a country with universal, publicly funded healthcare similar to Canada, demonstrated net savings for the health system and the potential for improved equity in access to care and health…”
Altomare, F., Kherani, A., & Lovshin, J. A. (2018) · Canadian Journal of Diabetes · doi:10.1016/j.jcjd.2017.10.027 · Cited by 42 · see citing works
Ming, S., Xie, K., Lei, X., Yang, Y., Zhao, Z., Li, S., Jin, X., & Lei, B. (2021) · International Ophthalmology · doi:10.1007/s10792-020-01685-x · Cited by 36 · see citing works
Feasibility and acceptance of artificial intelligence-based diabetic retinopathy screening in Rwanda
Whitestone, N., Nkurikiye, J., Patnaik, J. L., Jaccard, N., Lanouette, G., Cherwek, D. H., Congdon, N., & Mathenge, W. (2024) · British Journal of Ophthalmology · doi:10.1136/bjo-2022-322683 · Cited by 31 · see citing works
Self-citationBhambhwani, V., Whitestone, N., Patnaik, J. L., Ojeda, A., Scali, J., & Cherwek, D. H. (2025) · Ophthalmic Epidemiology · doi:10.1080/09286586.2024.2434738 · Cited by 11 · cited 2× · see citing works
“6 The results are paralleled in the modified base case that includes the retinal OCT screening, with the AI-based cost per diagnosed case being 53% lower than the traditional physician-based approach.”
Self-citationFuture trends in ophthalmology health human resources in Canada
Bellan, L. (2016) · Canadian Journal of Ophthalmology · doi:10.1016/j.jcjo.2016.03.001 · Cited by 9 · cited 1× · see citing works
“Given the projected shortages of ophthal-mologists 28 in Canada and long wait lists, 29 decreasing the number of patients requiring specialist time relieves additional health system challenges and has been shown to increase specialist productivity, 30 while also increasing adherence to recommended…”
Flagged first, then by citations
Research works printed but not linked: 2
- International Diabetes Federation . IDF diabetes atlas. 2021. Available: https://diabetesatlas.org/atlas/tenth-edition/ [Accessed 20 Aug 2024].
- Drummond MF , O’Brien B , Stoddart GL , et al . Methods for the Economic Evaluation of Health Care Programmes 2nd Edition. Oxford: Oxford University Press, 2000.
- Authors
- Di Matteo, L., Whitestone, N., & Bhambhwani, V.
- Affiliations
- Lakehead University, Thunder Bay, Ontario, Canada
- Orbis International, New York City, New York, USA
- Thunder Bay Regional Health Sciences Centre (CA)
- Venue
- BMJ Open Ophthalmology
- Publisher
- BMJ
- Published
- 2026-07
- Volume
- 11
- Issue
- 3
- Pages
- e002904
- ISSN
- 2397-3269 (ISSN-L)
- PMC
- PMC13422802
- OpenAlex
- W7171868654
- Semantic Scholar
- 8eb3f864737bc99c5ace8d6cafea3e5c79ced330
- Cited by
- 0
- References
- 20
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- BMJ Open Ophthalmology (publishedVersion, cc-by-nc)
- PubMed (publishedVersion)
- PubMed Central (submittedVersion, cc-by-nc)
- Funders
- Orbis Canada (N/A)
- Northern Ontario Academic Medicine Association (A-23-02)
- Eyenuk (N/A)
- iCare Inc (N/A)
- Topics
- Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Retinal Imaging and Analysis
- Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Health Systems, Economic Evaluations, Quality of Life
- Health Sciences › Medicine › Ophthalmology › Ophthalmology and Visual Health Research
- Keywords
- diabetic retinopathy, artificial intelligence, retinopathy screening, primary care, cost analysis, diabetes mellitus, indirect costs
- MeSH terms
- Adult, Aged, Artificial Intelligence, Costs and Cost Analysis, Diabetic Retinopathy, Female, Humans, Male, Mass Screening, Middle Aged, Primary Health Care, Health Care Costs
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-13.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-13.
Cited sources are checked by gpt-6-sol.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
