AI tools are already triaging suspected skin cancer patients in parts of the NHS, with early results showing real promise — but also real limits. A new UK consortium, SkinRAI, is stepping in to set standards, close evidence gaps, and ensure human oversight stays in the loop. The technology is moving fast; the frameworks to govern it are still catching up.
AI is already being used to triage suspected skin cancer patients in parts of the UK's NHS, and early real-world data are offering a mixed but encouraging picture. The DERM AI tool — a CE-marked medical device that analyzes dermatoscopic images — allowed 22.3% of patients to be discharged before seeing a dermatologist in its first two months at one NHS trust. At another site, nearly 4,700 referrals have been processed, with significant reductions in unnecessary face-to-face consultations.
But the technology isn't without its gaps. DERM was highly sensitive (catching 95% of cancers vs. 88.5% for dermatologists), yet less specific and less accurate at identifying the correct diagnosis. Crucially, four cancers were found in lesions the AI had flagged as benign — underscoring why human review remains essential.
To address growing concerns about oversight, the British Association of Dermatologists has launched the Skin and Responsible AI (SkinRAI) Consortium, a clinician-led initiative focused on setting standards for how AI is developed, evaluated, and monitored in dermatology.
Key Takeaways:
Why it matters: As AI tools move from pilots to routine care, the stakes for getting governance right are high. SkinRAI's work could serve as a model for how other specialties responsibly integrate AI — ensuring innovation doesn't outpace patient safety.
