Abstract
Purpose :
The cost-effectiveness of AI glaucoma screening in the West is unknown. We use comprehensive and dynamic decision-analytic Markov models to offer a robust estimation of the cost-effectiveness and cost-utility of AI-assisted population screening for glaucoma versus traditional screening in Australia.
Methods :
The Markov model simulated glaucoma progression in a cohort of 12,283,846 Australians aged 50+ in 2021 through 40 one-year cycles. It encompassed five disease stages: glaucoma unlikely, glaucoma suspect, glaucoma certain, blindness, and death. Four scenarios were studied: status quo and three interventions using AI-assisted screening. Status quo involved 23.216% of the population undergoing optometry exams for glaucoma. Scenario A replaced manual screening with AI, Scenario B augmented AI screening based on a 62% surveyed-based patient satisfaction rate on AI, and Scenario C entailed universal AI screening for Medicare-eligible individuals. Costs for screening, consultations, and treatment were sourced from various studies and benefit schemes as of July 2023. The cost of AI-assisted screening considered patent, hardware, depreciation, and labour expenses associated with our developed AI products.
Results :
In the status quo, the cohort of individuals in Australia was projected to accumulate 177,469,478 quality-adjusted life years (QALYs) over 40 years, with a total care cost of $40,640.3 million. Scenario A was cost-saving and provided the cohort with an additional 7,454 QALYs over 40 years compared to the status quo. The healthcare cost projection for this scenario was $39,536.2 million. The benefit-cost ratio stood at 18.5, resulting in a net monetary benefit (NMB) of $1,476.8 million. Scenario B increased the health benefit with an additional 110,234 QALYs gained over 40 years compared to the status quo. The projected healthcare cost for this scenario was $38,236.0 million. The benefit-cost ratio was 2.5, yielding a NMB of $7,916.0 million. Scenario C, a universal AI-based screening, provided an additional 126,201 QALYs over 40 years compared to the status quo, with a healthcare cost of $38,005.2 million. The benefit-cost ratio was 2.3, resulting in a NMB of $8,945.2 million.
Conclusions :
AI glaucoma screening is more cost-effective than traditional glaucoma screening in all modelled scenarios.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.