Description

Background Diabetic retinopathy (DR) and age-related macular degeneration (AMD) are major cause of vision impairment. With an aging population and the increasing prevalence of diabetes, the incidence of both DR and AMD continue to rise. However, due to limited disease literacy and lack of dedicated fundus screening tools in department of internal medicine, many patients are diagnosed and treated at a late stage of the disease. This study aims to evaluate the clinical value of VeriSee artificial intelligence (AI) - assisted diagnostic software among patients with diabetes and the elderly population, focusing on their screening effectiveness and feasibility.

Objective This study aims to evaluate the effectiveness of the VeriSee - an AI-assisted diagnostic software for DR and AMD - in improving the screening rates of macular degeneration, diabetic retinopathy and glaucoma, as well as reducing the incidence of severe visual impairment and lowering overall healthcare burdens. Simultaneously, the investigators will conduct and cost-effectiveness assessment of the VeriSee AI-assisted diagnostic software.

Methods This study is a multicenter, two-arm, parallel-group, open-label, individual-level randomized controlled trial (RCT) conducted at the main branch and Bei-Hu branch of National Taiwan University Hospital. Study participants include: (1) individuals aged 50 and above who meet the screening criteria for AMD; and (2) individuals aged 20 and above with diabetes who meet the screening criteria for DR. Participants are randomized into two groups: (1) the intervention group (AI-assisted screening) in which participants will receive the AI-assisted image analysis followed by immediate explanation of results by a physicians, with ophthalmology referral as needed; and (2) the control group (physician only screening), in which participants undergo standard fundus photography interpreted by physicians, with results discussed during a subsequent visit. During the trial, the ophthalmology referral rates and subsequent diagnostic outcomes will be tracked to evaluate the effectiveness of the AI-assisted diagnostic approach.

Results The study was funded in September 2024. Data collection is expected to last from April 2025 to December 2027. The primary outcome of this study is the detection rate of DR and AMD using the AI-assisted diagnostic software and its impact on diagnosis and treatment following the referral. Referral outcomes will be tracked through electronic medical records (EMR), and both patient and physician satisfaction survey will be conducted to evaluate the feasibility and acceptability of AI implementation in clinical settings.

Conclusions This study is expected to provide evidence on the clinical effectiveness and application value of AI-assisted ophthalmic screening, while also exploring its impact on healthcare procedures and patient care. By enhancing the detection rate of retinal diseases among individuals with diabetes and the elderly, AI-assisted technologies may facilitate earlier diagnosis and timely treatment, potentially improving the visual health and overall quality of life.