医学教育管理 ›› 2026, Vol. 12 ›› Issue (5): 679-686.doi: 10.3969/j.issn.2096-045X.2026.05.016

• 调查研究 • 上一篇    下一篇

数智时代下卓越眼科医学人才培养的伦理风险及教育治理

  

  1. 1.同济大学医学院,上海 200092; 2.同济大学附属同济医院眼科,上海 200065; 3.同济大学医学院视觉科学与转化研究中心,上海 200065
  • 收稿日期:2026-02-27 修回日期:2026-03-20 出版日期:2026-11-13 发布日期:2026-11-13

Ethical risks and educational regulation in the cultivation of outstanding ophthalmology talents in the digital-intelligent era

  1. 1. School of Medicine, Tongji University, Shanghai 200092, China; 2. Department of Ophthalmology, Tongji Hospital Affiliated to Tongji University, Shanghai 200065, China; 3.Center for Vision Science and Translational Research, School of Medicine, Tongji University, Shanghai 200065, China
  • Received:2026-02-27 Revised:2026-03-20 Online:2026-11-13 Published:2026-11-13

摘要:

目的 分析数智时代卓越眼科医学人才培养中的核心伦理风险,并探索教育干预策略,为医学人才伦理素养培养提供实证参考。方法 ①文献分析与政策梳理:系统梳理全球眼科(尤其是中国)数智医疗伦理相关研究;②案例深描:解构同济大学医学院眼科学临床教学、科研训练等4类场景的伦理风险及联邦学习、区块链等技术方案的伦理适配性;③相关者调研:选取同济大学医学院2024—2025年在读眼科学研究生、规培学员68人,教师18人,临床医师 22人,患者127人开展问卷与访谈;④教育干预试验:将68名学生随机分为试验组(n=34,接受场景化伦理教育)与对照组(n=34,接受传统伦理教学),对比干预效果。结果 共梳理出数据伦理、AI 应用伦理等6类23项风险问题,其中联邦学习在跨机构数据共享中存在权限界定争议,区块链技术应用存在可解释性不足问题;调研显示82.4%(56/68)的学生对数据隐私伦理认知不足,66.7%(12/18)的教师缺乏数智伦理教学能力;干预后试验组伦理认知得分(89.6±4.2)分,显著高于对照组(75.3±5.8)分(t=11.26,P<0.001),伦理决策案例通过率(88.2%)高于对照组(55.9%)(χ2=8.838,P=0.003)。结论 数智时代医学人才培养伦理风险多元且复杂,“场景嵌入、能力导向、多元协同”的教育管控策略可有效提升学生伦理素养,可为医学院校开展数智伦理教育提供参考。

关键词:

"> 数智时代|卓越医学人才|医学伦理|眼科学|教育管控|人工智能|医学教育|临床胜任力

Abstract:

Objective To systematically analyze core ethical risks in cultivating clinically competent ophthalmologists and to design effective educational interventions in the digital intelligence era, and to provide robust empirical evidence for ethical literacy development and curriculum innovation in medical education. Methods  First, a systematic scoping review of global ophthalmic digital health ethics literature and policies was conducted, with particular emphasis on the Chinese context and regulatory framework. Second, in-depth qualitative case studies were performed to deconstruct ethical risks across four critical teaching scenarios (clinical instruction, research training, telemedicine practice, and artificial intelligence-assisted diagnosis) and to rigorously evaluate the ethical fitness and implementation challenges of emerging technological solutions including federated learning and blockchain infrastructure. Third, a multi-stakeholder investigation was undertaken comprising 68 postgraduate students and residents, 18 faculty members, 22 practicing clinicians, and 127 patients from 2024 to 2025 through validated structured questionnaires and semi-structured in-depth interviews. Fourth, a prospective randomized controlled educational intervention trial was conducted, in which 68 students were randomized into an experimental group (n=34, receiving immersive scenario-embedded ethics education with simulation-based training) and a control group (n=34, receiving conventional didactic ethics instruction), with primary and secondary outcomes compared between groups using appropriate statistical methods. Results  A comprehensive framework of twenty-three risk items across six interconnected domains was identified and categorized, encompassing data ethics, artificial intelligence application ethics, algorithmic bias and fairness, informed consent in digital contexts, transformation of professional-patient relationship in telemedicine, and research integrity in big data environments. Specifically, federated learning demonstrated significant authority-boundary controversies and accountability gaps in cross-institutional data sharing, while blockchain technology exhibited critical limitations in algorithmic explainability, transparency, and regulatory compliance. Stakeholder surveys revealed that 82.4% (56/68) of students exhibited inadequate awareness of data privacy ethics and security protocols, and 66.7% (12/18) of faculty members reported deficient pedagogical competency and confidence in teaching digital ethics. Following the twelve-week educational intervention, the experimental group exhibited significantly superior performance compared with the control group in validated ethical cognition scores (89.6±4.2 vs 75.3±5.8, t=11.26, P<0.001) and objective structured ethical decision-making case clearance rates (88.2% vs 55.9%, χ2=8.838, P=0.003). Conclusion  Ethical risks in medical talent cultivation are multifaceted, dynamic, and increasingly complex in the digital intelligence era. The scenario-embedded, competency-oriented, multi-stakeholder collaborative governance strategy effectively enhances students' ethical literacy, critical reasoning, and decision-making capabilities, offering a replicable, scalable, and context-adaptable paradigm designated for cultivating outstanding medical professionals in ophthalmology and potentially across broader medical and surgical disciplines.