Bradford NHS clinic sees a third more skin cancer patients after deploying AI triage tool

A dermatology clinic at St Luke's Hospital in Bradford is seeing eight more patients per session than it did a year ago — not by hiring more doctors, but by letting an AI system sort the urgent from the routine before a clinician ever picks up a scalpel.

Bradford Teaching Hospitals NHS Foundation Trust introduced DERM — Deep Ensemble for the Recognition of Malignancy, developed by Skin Analytics — in April, making it the first trust in West Yorkshire to adopt the technology. The system is now used in a new thrice-weekly teledermatology service at St Luke's, and early figures show the clinic has increased capacity from 24 to 32 patients per session while reducing unnecessary biopsies by around 10%.

The workflow is straightforward: a healthcare assistant photographs a suspicious lesion using approved imaging equipment and submits the images to DERM. The system analyses them and recommends either discharge or referral for further assessment. Crucially, every AI recommendation is then reviewed by a clinician — the technology is positioned as a decision-support layer, not an autonomous diagnostician.

Combining this cutting-edge AI will give us the capacity to pick up potentially serious skin lesions quicker and speedier than current processes.— Zakir Shariff, Consultant Plastic Surgeon and Clinical Lead for skin cancer

Shariff called DERM 'the future of skin cancer diagnosis in this country.' Around 5,000 suspected skin cancer referrals arrive at Bradford Teaching Hospitals annually, yet confirmed malignancies account for only about 8 percent of those cases — approximately 400 patients. That gap means the vast majority of referrals are consuming specialist time for conditions that turn out to be benign, creating long waits for the minority who genuinely need urgent care.

The same pressure exists across the UK: only 6% of urgent skin cancer referrals from primary care nationally are confirmed as malignant cancer cases, according to government-transformation.com, making efficient triage a system-wide challenge.

DERM addresses this by routing benign cases to non-urgent pathways and discharging patients who do not need further intervention, while flagging suspicious lesions for immediate escalation to a consultant dermatologist — at St Luke's, one is based next door to the teledermatology service. Tissue from those cases is sent to a laboratory for definitive analysis.

For many patients, weeks of biopsies and awaiting results can be avoided as the process will reduce prolonged uncertainty and help avoid more invasive surgery.— Tom White, General Manager for the Musculoskeletal and Therapies Clinical Support Unit

White added that the service may eventually be rolled out in community settings and GP surgeries, sparing patients a hospital visit altogether.

Laurence Patten, 73, who has previously had skin cancer and also lives with multiple myeloma, was sent for an urgent referral after his GP spotted a mole on his back that looked suspicious. After DERM identified the lesion as a concern, a specialist reviewed it and concluded it was probably benign. The lesion is being removed and tested to confirm a definitive diagnosis, illustrating how the AI and clinical judgment work in tandem rather than in sequence.

DERM is already in use at around 25 other NHS trusts across the UK. According to Skin Analytics Performance Reports, cited by government-transformation.com, the system is 99.7% accurate at ruling out skin cancers. Bradford has taken on the tool as part of a three-year programme, with the Cancer Alliance set to examine the trust's first-year data for evidence of both safety and effectiveness.

NICE has recommended DERM for NHS teledermatology services during an evidence-generation period, while real-world data continues to be gathered to determine its longer-term clinical value. NICE classifies DERM as a UKCA Class IIa medical device intended for screening, triage and assessment of suspicious skin lesions in adults, not for definitive diagnosis. The regulator has also flagged the need for more evidence on how the technology performs across different skin tones, and notes it is not suitable for certain lesion types, including those on nails, mucosal surfaces, palms and soles, or lesions that are open, ulcerated, or previously biopsied.

For now, Bradford's early experience points to a model that other overstretched dermatology departments may find worth watching: AI handling the initial sort, clinicians handling the judgment calls, and patients — particularly those with genuinely dangerous lesions — waiting less.

Why it matters — With NHS dermatology departments under sustained pressure from rising referrals, Bradford's early results offer a concrete, numbered case for how AI triage tools — used alongside, not instead of, clinicians — could help more patients be seen faster without adding medical staff.

Reported by ceoindiamagazine.com, government-transformation.com, time.news