Development, validation, and cost-effectiveness analysis of an AI-assisted three-tiered glaucoma screening model in a...

Glaucoma is the leading irreversible blinding eye disease worldwide and the global prevalence of glaucoma for individuals aged 40-80 years is estimated as 3.54%. Early screening and prevention wherever possible are essential interventions of chronic disease management for glaucoma, but there is currently a lack of a recognized glaucoma screening model for Chinese population. Therefore, this study intends to construct and validate an artificial intelligence (AI)-assisted three-tiered glaucoma screening model based on the community population and assess its cost-effectiveness.

This is a community population-based cluster randomized controlled trial with a minimum of 6-year follow-up. A three-tiered glaucoma screening strategy appropriate for Chinese individuals will be established, including a high-risk model questionnaire (primary screening), ophthalmic examinations (secondary screening, combined with physician-based and AI-assisted image interpretation), and definitive diagnosis by tertiary hospitals (tertiary screening). The participants of each community will be randomly divided into three groups (simple cluster randomization): no screening group, routine screening group, and tiered screening group. Participants in the no screening group will receive regular glaucoma-related health education, structured annual follow-up for symptom monitoring and recording of external ophthalmology visits, and a comprehensive ophthalmic screening at the end of the study to compensate for the absence of regular screening. A total of 28,275 participants (9,425 per group) will be recruited, allowing for an anticipated loss-to-follow-up rate of 20%. All primary outcome measures will be analyzed on a per-participant basis. The main endpoint is the glaucoma detection rate, with sensitivity and specificity of the screening as additional primary outcomes. Secondary outcomes involve comparative analyses of the cost-effectiveness of different screening strategies and the diagnostic accuracy of physician-based versus AI-assisted image interpretation.

The protocol has been approved by the Biomedical Ethics Review Committee, West China Hospital of Sichuan University [2025(1021)]. All the participants will be required to afford signed informed consent. The study results will be presented at scientific meetings and published in a peer-reviewed journal.

https://www.chictr.org.cn/showproj.html?proj=281642, identifier (ChiCTR2500107852).