A surge in artificial intelligence tools for skin cancer detection is offering new hope in tackling one of the UK’s fastest-rising cancers, particularly for those with underrepresented skin tones.

Record NHS referrals highlight urgent need

In 2023/24, more than 650,000 people in England were referred onto the NHS’ urgent suspected skin cancer pathway, a record high. A new AI tool aims to provide faster, more accurate melanoma diagnoses.

Searches for “AI skin cancer detection” have risen by more than 5,000% in the past 30 days, reflecting growing interest in technology-led dermatology. One system leading the way is DERM, an AI tool that analyses images of moles to triage patients and help clinicians prioritise urgent cases. So far, it has achieved 99.9% accuracy in ruling out melanoma and identified over 15,000 skin cancers.

Rising melanoma rates and preventability

Melanoma is the fifth most deadly cancer in the UK, with incidence rates rising nearly 31% over the past decade. Last year, 20,800 new cases were recorded, with more than six deaths daily attributed to the disease. However, experts believe approximately 86% of melanoma cases in the UK are preventable.

“AI has the potential to bridge long-standing diagnostic gaps in dermatology,” says Dr Conal Perrett, consultant dermatologist at The Devonshire Clinic. “By training algorithms on more diverse skin types, we can ensure that early warning signs of melanoma are recognised equally, regardless of skin tone or geography. For too long, the diagnostic images used to train dermatology AI systems have under-represented darker skin. That’s beginning to change, and it’s an important step toward health equity.”

Addressing disparities with inclusive AI

Traditional training datasets for skin cancer algorithms have focused on lighter skin tones, leading to lower accuracy for people with darker complexions. Studies from Cancer Research UK show this can delay diagnoses in ethnic minority groups, where melanoma may appear on the palms or soles. By incorporating broader datasets, tools like DERM aim to improve precision across all demographics, supporting the NHS’s goal of equitable care.

Integrating AI into teledermatology also aligns with the NHS’s wider focus on digital transformation, reducing wait times and geographical barriers to specialist care.

“The most valuable use of AI is as an assistive tool, helping dermatologists assess lesions more efficiently, reduce diagnostic variation, and triage patients more effectively in remote consultations,” Dr Perrett explains. “That’s why in-person mole screenings and consultant-led dermatology services are still extremely vital.”

DERM system performance

Developed by Skin Analytics and recently granted conditional approval by NICE, DERM demonstrated 100% sensitivity in detecting malignant cases in a prospective trial involving over 1,500 lesions across seven UK hospitals, with an AUROC of 95.8%. Its performance matches or surpasses clinical specialists, potentially halving unnecessary referrals while prioritising high-risk patients.

Additional preventative measures

Research continues into preventative measures such as vitamin B3 (nicotinamide) and non-invasive diagnostic patches. Nicotinamide has reduced non-melanoma skin cancer risk by up to 23% in high-risk individuals, according to the New England Journal of Medicine. Emerging diagnostic patches use biosensors to monitor UV exposure and skin changes in real time, alerting users to potential issues before they escalate.

“Ultimately, early detection is still our most powerful tool against melanoma,” adds Dr Perrett. “Whether through AI-driven image analysis, smart wearable sensors, or routine skin checks, recognising changes early can make all the difference.”

Top 5 signs your mole could be a concern

• Uneven shape or border: Asymmetrical or jagged edges may need closer inspection.
• Colour changes: Multiple shades of brown, black, red, or blue can signal concern.
• Growth or evolution: Rapidly changing moles or ones that itch, bleed, or crust over should be assessed.
• The “odd one out”: A mole that looks different from your others may indicate risk.
• New or unusual marks: On darker skin, melanoma may appear on palms, soles, or under nails.

These signs are often remembered by the ABCDE rule (Asymmetry, Border irregularity, Colour variation, Diameter over 6mm, Evolving changes). Public health campaigns emphasise that melanoma survival rates exceed 90% when caught early.

“If something looks or feels different, don’t ignore it,” says Dr Perrett. “Most skin changes are harmless, but when caught early, melanoma is highly treatable, so trust your instincts and get it checked.”

AI as a complement, not a replacement

As AI tools like DERM integrate into routine practice, they promise to democratise access to expert-level diagnostics, particularly in underserved areas. Experts stress that technology complements, rather than replaces, human vigilance and professional oversight. Combining tech innovation with sun safety education could significantly reduce preventable cases.