Artificial intelligence (AI) technologies for assessing and triaging skin lesions within the urgent suspected skin cancer pathway: early value assessment

DERM is an artificial intelligence (AI)-based skin lesion analysis software technology intended for use in the screening, triage and assessment of suspected skin cancer lesions, in people aged 18 or over. It is intended to be used as an automated tool or with a second read, to decide if further assessment by a dermatologist is needed. A smartphone is used to capture images of skin lesions using a dermoscopic lens attachment, and the images are uploaded to the online platform. The DERM platform uses an AI-based fixed algorithm (it does not update itself automatically) to analyse the dermoscopic images and provide a suspected diagnosis of the lesion. If DERM labels the lesion as benign, the person is discharged from the urgent suspected skin cancer pathway and is told the results with safety netting advice. If DERM labels the lesion as pre-cancer or cancer, an NHS dermatologist reviews the case virtually and decides on a management plan for the person. DERM can classify lesions as: melanoma, squamous cell carcinoma, basal cell carcinoma, intra-epidermal carcinoma, actinic keratosis, atypical nevus or benign lesions (this includes benign vascular lesion, seborrheic keratosis, dermatofibroma, solar lentigo and melanocytic benign nevus). If a lesion has features of more than 1 lesion type, DERM uses a risk hierarchy to diagnose the lesion as the more severe suspected lesion type. DERM is used within teledermatology services after referral from primary care. The cost of using the online platform for a DERM assessment is £30 per referral. There is an extra cost of £8.20 per referral if NHS teledermatology staff virtually review a case to decide on the most appropriate outcome. The total price can be discounted to £35.90 if the subsequent biopsy results from the lesions that have been assessed by the technology are shared with the company. It costs an extra £17 to have a case reviewed by the company's second-read dermatologist. The company state that these costs include training and data storage. DERM is UKCA marked (class IIa) for AI used as a medical device.

Moleanalyzer pro is a software, intended to be used by a healthcare professional for assessing single skin lesions to help to recognise melanoma lesions. The technology is not intended to be used to confirm a clinical diagnosis of melanoma and can be used for any age group. The target population is people with skin lesions, moles or multiple nevus syndrome. Moleanalyzer pro is used with the FotoFinder Universe software platform which includes the FotoFinder AI scoring assistant. FotoFinder provides 2 options: online AI in which the algorithm is updated continuously and offline AI in which the algorithm can be updated annually. The software needs a dermoscopic image of the lesion for the AI score analysis. The AI score is based on comparisons with images of cancer skin tumours, such as melanoma, basal cell carcinoma, lentigo maligna, squamous cell carcinoma, actinic keratosis and many others. The score indicates how similar a lesion is to these comparison images, so it only provides a statistical estimate of the similarity of the person's lesion to the cancer lesion images. A score of 0 to 0.2 indicates the lesion is inconspicuous, 0.21 to 0.49 indicates further clarification is needed, and 0.50 to 1.0 indicates a conspicuous lesion which should be observed with great attention. The cost of the FotoFinder AI scoring assistant is offered at a flat fee of £1,210 plus VAT per year for single-user access. The Moleanalyzer pro including AI scoring assistant's offline package is £1,750 plus VAT per year for single-user access. There is no cost for training and there is a discount available for multi-user access. Moleanalyzer pro is CE-marked (class IIa) for AI used as a medical device.

Skin cancer is an abnormal growth of skin cells and most often develops on skin that has been exposed to the sun. There are 3 major types of skin cancer: melanoma, squamous cell carcinoma (SCC), and basal cell carcinoma (BCC). There are also other rare skin cancers. The first assessment of a skin condition is done by a GP in primary care, to determine the appropriate referral pathway. Section 1.7 on skin cancers in NICE's guideline on suspected cancer describes the criteria for urgent referral of suspected skin cancer. People whose lesions are referred on this pathway should be given a diagnosis or ruling out of cancer within 28 days of being referred by their GP.

Historically, GPs directly referred everyone with suspicious skin lesions for a face-to-face appointment with a consultant dermatologist, using the urgent suspected skin cancer pathway. Face-to-face appointments are still used for people with multiple suspicious lesions, a history of skin cancer or other risk factors, or when other clinical pathways are unavailable in the local area or are unsuitable.

NHS dermatology services need to reduce backlogs and delays in providing face-to-face appointments because of limited staff, including dermatologists. The demand from an increasing number of urgent referrals for suspected skin cancers can mean that face-to-face appointments for people with other non-cancer skin conditions (including painful and debilitating inflammatory skin conditions) are delayed. The NHS Plan introduced teledermatology services for triaging, diagnosing, and managing skin conditions without the person being physically present. These services are intended to help manage demand and reduce the number of face-to-face appointments offered to people with low-risk non-cancer lesions. This evaluation assesses the benefits of using AI technologies within teledermatology services.

The 2 comparators are teledermatology alone and face-to-face dermatology assessment.