AI-assisted glaucoma screening in Singapore's diabetic eye program

A Pragmatic Randomized Controlled Trial of a New Artificial Intelligence-Assisted Clinical Model in Opportunistic Screening for Glaucoma in the Singapore Integrated Diabetic Retinopathy Program

It will test whether an AI algorithm can detect glaucoma from color fundus photos in adults with diabetes attending Singapore's diabetic eye screening program.

Quick facts

• Phase: NA
• Study type: Interventional
• Enrollment: 1040 (estimated)
• Ages: 21 Years and up
• Sex: All
• Sponsor: Singapore Eye Research Institute (other)
• Locations: 1 site (Singapore, Singapore)
• Trial ID: NCT07243665 on ClinicalTrials.gov

What this trial studies

This pragmatic, single-blind randomized controlled trial will recruit 1,040 adults with diabetes from two Singapore centers and randomize them 1:1 to AI-assisted fundus-photo grading or current manual grader assessment. After fundus imaging, both arms' results will be compared against a gold-standard glaucoma diagnosis determined by comprehensive ophthalmic examination including intraocular pressure, visual fields, optical coherence tomography, and dilated fundus assessment. Diagnostic accuracy, referral decisions, and downstream costs will be compared, and a cohort-based Markov model will estimate lifetime costs and cost-effectiveness. The trial is embedded within the Singapore Integrated Diabetic Retinopathy Programme to test the algorithm in a real-world screening pathway.

Who should consider this trial

Good fit: Adults aged 21 and older with type 1 or type 2 diabetes who attend the participating SiDRP clinics in Singapore, can have retinal photos taken, and can provide informed consent are the intended participants.

Not a fit: Patients who cannot have usable retinal photographs, cannot complete the ocular examination protocols, or have contraindications as identified by their clinicians are unlikely to benefit from this screening approach.

Why it matters

Potential benefit: If successful, the AI tool could speed up and standardize glaucoma detection in diabetic screening, enabling earlier referrals and potentially reducing vision loss.

How similar studies have performed: Retrospective and validation studies using deep learning on fundus photos have shown promise for glaucoma detection, but prospective randomized trials in routine screening programs are still limited.

Eligibility criteria

Inclusion Criteria: We aim to recruit all eligible patients who attend Singapore General Hospital (SGH) Diabetes \& Metabolism Centre's (DMC) clinics and SingHealth Polyclinics (SHP)-Bukit Merah under the Singapore Integrated Diabetic Retinopathy Programme (SiDRP). Patients are eligible for the study if

1. Aged 21 years old and above, with diabetes, including type 1 and type 2,
2. Retinal photos of the patients can be taken with the fundus camera in the clinics, regardless of photos' quality, and
3. They are willing and capable of providing a written informed consent form.

Exclusion Criteria: Patients meeting any of the exclusion criteria will be excluded from participation:

1. Patients who have difficulty in having retinal photos taken or have difficulties in completing the ocular examination protocols according to investigator's decision.
2. Any other contraindication(s) as indicated by the endocrinologists responsible for the patients.

   \-

Where this trial is running

Singapore, Singapore

• Singapore National Eye Centre — Singapore, Singapore, Singapore (RECRUITING)

Study contacts

• Principal investigator: Ching-Yu Cheng, MD, PhD — Singapore Eye Research Institute
• Study coordinator: Ching-Yu Cheng, MD, PhD

Conditions: Glaucoma, deep learning, fundus photos, artificial intelligence, randomised controlled trial, screening