A dermatology clinic at St Luke’s Hospital in Bradford is seeing a third more patients after adopting an artificial intelligence system to assess suspicious skin lesions. Since the technology arrived in April, staff have been able to see 32 patients in each session, up from 24.
The system, called DERM, short for Deep Ensemble for the Recognition of Malignancy, is made by Skin Analytics, which says it is 99.9% accurate at ruling out melanoma. A healthcare assistant takes three photographs of a mole or lesion and uploads them, and the software returns an assessment within minutes. Patients whose cases look benign receive advice and are discharged, while those flagged as suspicious go on to see a specialist at a one-stop clinic. In Bradford, a clinician double-checks every image before a diagnosis is confirmed.
Consultant plastic surgeon Zakir Shariff said the system had eased bottlenecks across the service, from patient waiting times to tests that did not need to happen, and was particularly good at giving quick non-cancerous diagnoses. According to the hospital trust, the number of unnecessary biopsies has fallen by about 10% since the AI was introduced.
One recent patient, 73-year-old former builder Lawrence Patten, was referred urgently by his GP over a mole on his back. The AI flagged the images as suspicious, and although a specialist examination suggested the mole was likely benign, it will be removed and tested to confirm. Dr Nader Ghaderi, an associate specialist in dermatology, said the system was right to flag the case, since a doctor would also have ordered a biopsy or a follow-up. Patten said he was glad to be seen so quickly. “The sooner they find things, the quicker they can sort it,” he said.
The National Institute for Health and Care Excellence has recommended the technology for NHS use for the next three years while more evidence is gathered. It is now in use at 40 NHS hospital trusts, and Bradford Teaching Hospitals was the first in Yorkshire to adopt it. The trust receives around 5,000 referrals for suspected skin cancer each year, of which only about 8%, or 400 patients, turn out to have a malignant cancer. Tom White, the general manager who oversees dermatology, said early evidence suggested the AI could be a “game changer”, cutting the number of urgent referrals dermatologists must review and freeing them to focus on the most serious cases.
