Ran An Ran, Lui Chun Ho, Tham Yih-Chung, Cheng Ching-Yu, Lam Chiu Yu, Cheung Wai Lam, Chan Siu Ting, Ma Hok Ngai, Chow Raphael Walter L C, Yang Dawei, Tang Ziqi, Liu T Y Alvin, Tham Clement C, Cheung Carol Y
Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong SAR, China.
Lam Kin Chung, Jet King-Shing Ho Glaucoma Treatment and Research Centre, The Chinese University of Hong Kong, Hong Kong SAR, China.
Eye (Lond). 2025 Jun 13. doi: 10.1038/s41433-025-03878-z.
Thorough investigations of end-users' awareness, acceptance, and concerns about ophthalmic artificial intelligence (AI) are essential to ensure its successful implementation. We conducted a literature review on the acceptance of ophthalmic AI to provide an overall insight and qualitatively analysed the quality of eligible studies using a psychological model. We identified sixteen studies and evaluated these studies based on four primary factors (i.e., performance expectancy, effort expectancy, social influence, and facilitating conditions) and four regulating factors (i.e., gender, age, experiences, and voluntariness of use) of the psychological model. We found that most of the eligible studies only emphasized performance expectancy and effort expectancy, and in-depth discussions on the effects of social influence, facilitating conditions, and relevant regulating factors were relatively inadequate. The overall acceptance of ophthalmic AI among specific groups, such as patients with different eye diseases, experts in ophthalmology, professionals in other fields, and the general population, is high. Nevertheless, more well-designed qualitative studies with clear definitions of acceptance and using proper psychological models with larger sample sizes involving other representative and multidisciplinary stakeholders worldwide are still warranted. In addition, because of the multifarious concerns of AI, such as the economic burden, patient privacy, model safety, model trustworthiness, public awareness, and proper regulations over accountability issues, it is imperative to focus on evidence-based medicine, conduct high-quality randomized controlled trials, and promote patient education. Comprehensive clinician training, privacy-preserving technologies, and the issue of cost-effectiveness are also indispensable to address the above concerns and further propel the overall acceptance of ophthalmic AI.
深入调查终端用户对眼科人工智能（AI）的认知、接受程度和担忧，对于确保其成功实施至关重要。我们对眼科AI的接受情况进行了文献综述，以提供全面的见解，并使用一种心理模型对符合条件的研究质量进行了定性分析。我们确定了16项研究，并根据心理模型的四个主要因素（即绩效期望、努力期望、社会影响和便利条件）以及四个调节因素（即性别、年龄、经验和使用的自愿性）对这些研究进行了评估。我们发现，大多数符合条件的研究仅强调绩效期望和努力期望，而对社会影响、便利条件及相关调节因素的影响进行的深入讨论相对不足。眼科AI在特定群体中的总体接受度较高，这些群体包括患有不同眼病的患者、眼科专家、其他领域的专业人员以及普通大众。然而，仍需要更多设计良好的定性研究，明确接受度的定义，并使用适当的心理模型，涉及全球范围内更多具有代表性和多学科的利益相关者，样本量更大。此外，由于对AI存在诸多担忧，如经济负担、患者隐私、模型安全性、模型可信度、公众认知以及对问责问题的适当监管，因此必须注重循证医学，开展高质量的随机对照试验，并加强患者教育。全面的临床医生培训、隐私保护技术以及成本效益问题对于解决上述担忧并进一步推动眼科AI的总体接受度也不可或缺。
