The scale of the pressure on dermatology services provides the backdrop for the findings. Urgent suspected skin cancer referrals in England have almost tripled since 2009, yet only around 6% of those referrals result in an urgent skin cancer diagnosis. Meanwhile, approximately one in four dermatologist roles in the UK remains unfilled, creating a widening imbalance between demand and specialist capacity. Patients with benign lesions therefore often queue alongside those with aggressive malignancies, and the system strains to triage them efficiently. The new study tested whether a CE-marked Class III AI medical device could safely absorb a meaningful share of that low-risk workload without compromising patient safety.

The autonomous AI-supported pathway was introduced across two hospital sites following an initial validation period. Patients referred on urgent suspected skin cancer pathways were offered the option of AI-based assessment, and uptake was striking: 86% consented to autonomous decision-making. In total, the pathway managed 8,391 patients, representing 94% of all urgent suspected skin cancer referrals across the two hospitals. Using clinical photographs and dermoscopic images captured with smartphone-based technology, the system classified skin lesions and autonomously discharged benign cases, while routing higher-risk lesions to teledermatologists for specialist review.

The proportions of patients managed without any clinician involvement were substantial. After exclusions, the AI autonomously discharged 31% of patients at one hospital and 25% at the other without a dermatologist ever reviewing their case. Teledermatologists subsequently discharged a further 24% and 25% of patients respectively, meaning that in both settings roughly half of all referred patients left the pathway without a face-to-face specialist consultation. For a service in which one in four consultant posts is vacant, that shift in workload distribution carries obvious operational significance.

The efficiency gains extended beyond simple discharge rates. Compared with standard teledermatology, the autonomous pathway reduced the proportion of patients requiring routine follow-up from 27% to 12%. Biopsy rates also fell markedly, to 27% under the autonomous pathway compared with 43% for conventional face-to-face care, suggesting the system was not simply deflecting work but sharpening the selection of lesions that genuinely warranted invasive investigation. Fewer unnecessary biopsies mean fewer procedures, less patient anxiety and lower costs across the pathway.

Aggregated across the 16-month study period, the researchers estimated that the autonomous pathway saved 2,851 hours of clinician time relative to a traditional face-to-face model, an approximate 62% gain in clinical capacity. Based on standard 20-minute consultations, that equates to more than 8,500 additional face-to-face appointments across the two hospitals. Dr Lucy Thomas, the lead author and a consultant dermatologist at Chelsea & Westminster Hospital NHS Foundation Trust, argued that the true value of the technology lies in what it frees specialists to do. “We believe the greatest value of autonomous AI lies not in the technology itself, but in the specialist capacity it unlocks,” she said. “Every hour saved reviewing low-risk lesions can be reinvested in patients with skin cancer, helping them access timely treatment to improve prognosis, and in patients with severe inflammatory skin disease, where earlier access to specialist care and effective treatments can transform quality of life.”

Safety monitoring was a central component of the deployment, and the results offer some reassurance alongside important caveats. In a national dataset incorporating both study sites, the system’s sensitivity exceeded 98% for invasive melanoma, squamous cell carcinoma and basal cell carcinoma, with a specificity of 72.1%. High sensitivity is the critical metric in a cancer pathway, since missed malignancies carry the gravest consequences. Nevertheless, surveillance identified six false-negative cases that had been discharged by the pathway: five basal cell carcinomas and one melanoma in situ. These were detected through post-market surveillance, and no adverse outcomes were identified within the available follow-up period, but the figures underline that autonomous discharge is not error-free.

That experience shaped the study team’s central lesson about deploying medical AI responsibly. “One of the key lessons for us is that deploying an AI system safely isn’t a one-off exercise,” Dr Thomas said. “You need to keep monitoring it, understand when things go wrong, learn from those cases and make sure patients themselves know what to look out for.” The emphasis on continuous post-market surveillance reflects a broader shift in thinking about AI as a medical device: unlike a static drug with a fixed safety profile, an algorithm’s performance can drift with changing populations, imaging practices and referral patterns, demanding ongoing audit rather than a single approval milestone.

The study also highlights the practical mechanics of making such a pathway work at scale. Consent was sought from every patient offered autonomous assessment, and the high acceptance rate of 86% suggests that, when the process is explained clearly, patients are willing to trust algorithmic discharge decisions for low-risk lesions. The device itself, funded by Chelsea & Westminster Hospital NHS Foundation Trust and partly supported by a grant from La Roche-Posay, operates on images that can be captured with smartphone-based dermoscopy, lowering the infrastructure barrier for adoption in community settings and smaller clinics where dedicated hospital imaging equipment is unavailable.

The researchers are careful to frame the findings as a proof of concept rather than a universal prescription. “If these findings are replicated across larger populations and different healthcare settings, autonomous AI could become an important part of creating a more sustainable dermatology service – not by replacing dermatologists, but by allowing scarce specialist expertise to be focused where it can make the greatest difference to patients’ lives,” Dr Thomas concluded. With referral volumes still climbing and workforce shortages showing no sign of easing, the study offers one of the most concrete demonstrations to date that autonomous AI can operate safely inside a cancer pathway while returning thousands of hours of specialist time to the patients who need it most.

Subject of Research: Autonomous AI triage of urgent suspected skin cancer referrals in dermatology

Article Title: Autonomous AI could create capacity for thousands more dermatology appointments, real-world study finds

Article References: Autonomous AI could create capacity for thousands more dermatology appointments, real-world study finds. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: autonomous AI, dermatology, skin cancer, teledermatology, AI medical device, melanoma, basal cell carcinoma, NHS, clinical capacity, EADV Congress 2026, post-market surveillance, dermoscopy