Purpose: The rapid integration of machine learning in ophthalmology may exacerbate existing disparities in eye health among underserved populations. This study aimed to assess the representation and quality of reporting on equity-relevant sociodemographic characteristics in retinal studies utilizing artificial intelligence (AI).

Design: A systematic scoping review.

Methods: A comprehensive literature search was conducted using Ovid MEDLINE, Ovid Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL) from January 1, 2018, to July 29, 2023. The reporting of equity-relevant sociodemographic characteristics of participants was assessed using the PROGRESS equity framework.

Main outcome measures: The proportion of AI-based retinal studies reporting sociodemographic characteristics (e.g., race, ethnicity, gender, socioeconomic status), geographic diversity of dataset sources, and quality of reporting of racial and ethnic data.

Results: A total of 360 studies met inclusion criteria, the majority of which were retrospective (79.2%), imaging-based (99.4%) studies utilizing fundus photos (51.9%) or optical coherence tomography (43.1%) data from online databases (55.6%) or clinical charts (42.2%). Geographic diversity of studies was minimal, with only a few of the studies using clinical chart data that was collected in countries within the African (0.3%), Eastern Mediterranean (5.0%), and South-East Asian Region (6.7%). Sociodemographic reporting was limited, with 30.8% of studies reporting gender or sex, and only 8.9% reporting race or ethnicity of participants. Only 11.1% of studies tested AI algorithm performance by gender or sex, and 1.7% by race or ethnicity. The quality of racial and ethnic data reporting was poor overall, with significant gaps in how data was collected, determined, and classified.

Conclusions: AI-related retinal studies have limited equity considerations, particularly with regards to underrepresentation of diverse sociodemographic groups and insufficient racial and ethnic data reporting. Enhanced diversity in study populations and improved data collection and reporting methodologies are needed to ensure AI algorithms are trained and perform comparably across heterogenous populations affected by retinal diseases.

Keywords: Artificial intelligence; ethnicity; health equity; machine learning; ophthalmology; racial groups; systematic reviews.

Copyright © 2025. Published by Elsevier Inc.

目的： 在眼科迅速集成机器学习可能会加剧未得到充分服务的人群在眼部健康方面的现有差异。本研究旨在评估利用人工智能（AI）进行的视网膜研究中与公平性相关的社会人口特征的表现和报告质量。

设计： 系统综述性回顾。

方法： 使用Ovid MEDLINE、Ovid Embase和Cochrane中央对照试验注册库（CENTRAL）从2018年1月1日至2023年7月29日进行了全面的文献检索。使用PROGRESS公平框架评估了参与者与公平性相关的社会人口特征报告情况。

主要结果指标： 基于AI的视网膜研究中报告社会人口统计学特性（如种族、民族、性别、经济状况）的比例，数据集来源地的地理多样性以及种族和族裔数据报告的质量。

结果： 共有360项研究符合纳入标准，其中大多数是回顾性研究（79.2%），基于成像的研究（99.4%）使用眼底照片（51.9%）或光学相干断层扫描（OCT）数据（43.1%），这些数据主要来自在线数据库（55.6%）或临床病历记录（42.2%）。研究的地理多样性很小，只有少数研究使用的临床病历数据是在非洲地区（0.3%）、东地中海区域（5.0%）和东南亚区域（6.7%）收集的。社会人口统计学报告有限，其中仅30.8%的研究报告了性别或性别的信息，而只有8.9%的研究报告了参与者种族或民族的信息。仅有11.1%的研究测试了AI算法在性别或性别上的性能表现，且有1.7%的研究根据种族或族裔对算法性能进行了评估。总体而言，种族和族裔数据报告质量较差，在数据收集、确定和分类方面存在显著差距。

结论： 与AI相关的视网膜研究在公平性考虑上有限，特别是在代表性不足的社会人口统计学群体的代表性问题以及种族和族裔数据报告不足的问题上。为了确保AI算法能够在受视网膜疾病影响的不同人群中进行训练并表现一致，需要增强研究人群的多样性，并改进数据收集和报告方法。

关键词： 人工智能；民族性；健康公平；机器学习；眼科；种族群体；系统回顾。

版权 © 2025。Elsevier Inc. 出版。