Rahmati Masoud, Smith Lee, Piyasena Mapa Prabhath, Bowen Michael, Boyer Laurent, Fond Guillaume, Kazemi Abdolreza, Yon Dong Keon, Lee Hayeon, Sehmbi Tarnjit, Ahluwalia Sanjiv, Pardhan Shahina
CEReSS-Health Service Research and Quality of Life Center, Assistance Publique-Hopitaux de Marseille, Aix-Marseille University, Marseille, France.
CRSMP, Center for Mental Health and Psychiatry Research - PACA, Marseille, France.
Eye (Lond). 2025 May 30. doi: 10.1038/s41433-025-03849-4.
BACKGROUND/OBJECTIVES: Artificial intelligence (AI) assessment of diabetic retinopathy (DR) instead of scarce trained specialists could potentially increases the efficiency and accessibility of screening programs. This systematic review aims to systematically examine the uptake of follow-up appointments with initial computer-based AI and human graders of DR.
We conducted a systematic review and meta-analysis by screening articles in any languages in PubMed, MEDLINE (Ovid), EMBASE, Web of Science, Cochrane CENTRAL and CDSR published from database inception up to 20 August 2024. We used random-effects meta-analysis to pool the results as odds ratios (OR) with corresponding 95% confidence intervals (CI).
Data from a total of 20,108 patients with diabetes (6476 participants graded using AI and 13,632 participants graded by human-graders; age range of the participants 5 to 67 years) from six studies were included. The result of the pooled meta-analysis showed that initial AI assessment of DR significantly increased uptake of follow-up appointments compared to human grader-based (OR = 1.89, 95% CI 1.78-2.01, P = 0.00001).
The present systematic review and meta-analysis suggest that initial AI-based algorithm for screening DR is associated with an increased uptake of follow-up examination. This is most likely due to instant results being made available with AI based algorithms when compared to a delay in assessment with human graders.
背景/目的：使用人工智能（AI）而非稀缺的专业培训人员来评估糖尿病视网膜病变（DR），可能会提高筛查项目的效率和可及性。本系统评价旨在系统地研究最初基于计算机的AI和人类分级员对DR进行随访预约的情况。
我们通过筛选从数据库建立至2024年8月20日在PubMed、MEDLINE（Ovid）、EMBASE、Web of Science、Cochrane CENTRAL和CDSR上发表的任何语言的文章，进行了系统评价和荟萃分析。我们使用随机效应荟萃分析将结果汇总为比值比（OR）及相应的95%置信区间（CI）。
纳入了六项研究中总共20108例糖尿病患者的数据（6476名参与者使用AI分级，13632名参与者由人类分级员分级；参与者年龄范围为5至67岁）。汇总荟萃分析结果显示，与基于人类分级员的评估相比，DR的初始AI评估显著增加了随访预约的接受率（OR = 1.89，95% CI 1.78 - 2.01，P = 0.00001）。
本系统评价和荟萃分析表明，基于AI的DR筛查初始算法与随访检查接受率的增加相关。这很可能是因为与人类分级员评估的延迟相比，基于AI的算法能即时提供结果。
