Exploring the role of artificial intelligence in skin cancer management

Artificial intelligence (AI), particularly convolutional neural networks, has demonstrated diagnostic performance in skin cancer detection with accuracy comparable with that of dermatologists. While interest in integrating AI into the skin cancer screening pathways in the UK is growing, there is limited evidence describing dermatologists’ perspectives on its optimal implementation. The aim of this study was to explore UK dermatologists’ perspectives on the optimal implementation of AI in skin cancer screening pathways. A cross-sectional study was conducted using an anonymized online questionnaire distributed to UK dermatologists and dermatology trainees between January and February 2025 through deanery mailing lists and dermatology professional societies. The data were analysed using Google Sheets and RStudio. Some responses were collected using Likert scales, while others were based on free text. Ninety-six completed responses were analysed. Dermato­logists expressed cautious optimism towards AI, particularly for diagnostic accuracy and decision support. Respondents were more likely to biopsy lesions flagged as malignant by AI (median Likert score 4, interquartile range 4–4) and supported AI use in skin cancer diagnosis (median 4, interquartile range 3.5–5). However, concerns persisted regarding moderate overall impact on practice, medicolegal liability and the potential to exacerbate health disparities. Current AI use was associated with increased perceived impact on clinical practice (odds ratio 2.80) and greater concern about disparities (odds ratio 2.97), but a lower likelihood of excision when AI suggested benign features (odds ratio 0.38). No association was found between attitudes and age, role or workplace setting. Qualitative themes emphasized AI as an adjunct rather than a replacement for clinical expertise. UK dermatologists demonstrate cautious support for AI in skin cancer management, favouring an augmented intelligence model that preserves human oversight. Clear governance, training frameworks and equity-­focused deployment strategies are essential to ensure safe and effective integration of AI into dermatology practice.

Abstract from Crossref