Artificial Intelligence in Dermatology: Diagnostic Accuracy, Limitations, and Future Perspectives
Year: 2026
Type: conference-paper
Abstract: Objectives The aim of this review is to evaluate the role of artificial intelligence in dermatology, with emphasis on diagnostic performance, current clinical applications, limitations to clinical adoption, and future perspectives. Methods A systematic review of 29 peer-reviewed studies published since 2020 was conducted to evaluate diagnostic accuracy, clinical applications, limitations, and future directions of artificial intelligence in dermatology. The review was performed in accordance with PRISMA guidelines to ensure methodological precision and transparency. Results Artificial intelligence (AI) is widely applied in dermatology to support diagnostic decision-making through analysis of clinical and dermoscopic images, particularly in melanoma and non-melanoma skin cancers [1-7]. Reported diagnostic accuracy in skin cancer detection ranges from 70% to 99,7%, with pooled performance around 90% and high sensitivity and specificity [8,9]. AI is also increasingly used in inflammatory and chronic skin diseases, including psoriasis, atopic dermatitis, and acne, where diagnostic accuracies exceed 89% and pooled sensitivity is above 90%, enabling standardized severity scoring and disease monitoring [10-14]. Despite these advances, clinical implementation remains limited by dataset bias, underrepresentation of darker skin tones, lack of real-world validation, regulatory uncertainty, ethical concerns, and workflow integration challenges [15- 18]. Emerging approaches emphasize AI as a clinical decision-support tool, with future development focusing on multimodal models, federated learning, teledermatology integration, and hybrid AI–dermatologist systems [19,20]. Conclusions Artificial intelligence demonstrates substantial potential to enhance dermatological diagnostics, disease monitoring, and clinical decision-making. However, addressing validation, integration and ethical challenges is essential to enable the safe and effective incorporation of AI into routine dermatological practice.
Author: Jazbutytė, Mikeila
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Keywords: artificial intelligence in dermatology, skin cancer diagnosis, melanoma, non-melanoma skin cancer, dermoscopy, psoriasis, atopic dermatitis, acne, diagnostic accuracy, clinical decision support, data bias, darker skin tones, federated learning, teledermatology, AI implementation challenges, ethical considerations
Open Access status: green
OpenAlex ID: W7160473436
