Nahm William J, Sohail Nayyab, Burshtein Joshua, Goldust Mohamad, Tsoukas Maria
New York University Grossman School of Medicine, New York, New York, USA.
University of Illinois College of Medicine Rockford, Rockford, Illinois, USA.
Int J Dermatol. 2025 Sep;64(9):1568-1583. doi: 10.1111/ijd.17847. Epub 2025 May 19.
This comprehensive review examines artificial intelligence (AI) applications in dermatology, approved by the United States (U.S.) Food and Drug Administration (FDA) and international organizations, evaluating their clinical implementation and impact on healthcare delivery. We identified fifteen regulatory-approved AI devices globally, including three FDA-approved systems in the U.S. The FDA-approved devices primarily focused on melanoma and skin cancer detection through specialized hardware, while international platforms emphasized broader applications, mobile accessibility, and condition-specific tools for managing various skin conditions. Beyond these specific tools, we analyzed how AI can enhance clinical dermatology through screening systems, diagnostic support, administrative automation, and practice optimization. AI's integration into medical education can provide immediate feedback, support resident training, and complement traditional instruction, while patient education applications can improve treatment adherence through personalized content delivery. While AI shows promise across these domains, successful implementation requires addressing challenges in representation disparities, data privacy, algorithmic fairness, and clinical workflow integration. Future development should focus on standardized validation protocols, diverse training sets, robust real-world studies, and comprehensive assessment of patient outcomes beyond traditional performance metrics. AI's role appears most effective as augmentation to clinical expertise, particularly in improving access to specialized care and supporting clinical decision-making.
本综述探讨了美国食品药品监督管理局（FDA）和国际组织批准的人工智能（AI）在皮肤病学中的应用，评估了其临床应用情况以及对医疗服务的影响。我们在全球范围内确定了15种获得监管批准的AI设备，其中包括美国FDA批准的3种系统。FDA批准的设备主要通过专用硬件专注于黑色素瘤和皮肤癌检测，而国际平台则强调更广泛的应用、移动可及性以及用于管理各种皮肤疾病的特定病症工具。除了这些特定工具外，我们还分析了AI如何通过筛查系统、诊断支持、行政自动化和实践优化来增强临床皮肤病学。AI融入医学教育可以提供即时反馈、支持住院医师培训并补充传统教学，而患者教育应用程序可以通过个性化内容传递来提高治疗依从性。虽然AI在这些领域显示出前景，但成功实施需要应对代表性差异、数据隐私、算法公平性和临床工作流程整合等挑战。未来的发展应侧重于标准化验证协议、多样化训练集、强有力的真实世界研究以及超越传统性能指标的患者结局综合评估。AI作为临床专业知识的补充，其作用似乎最为有效，特别是在改善获得专科护理的机会和支持临床决策方面。
