Summary

Advice

The technologies described in this briefing are used for triaging, reporting and identifying possible melanoma in skin lesion images and include nomela, DERM, Moleanalyzer pro and SkinVision.
The innovative aspects are the signal processing and artificial intelligence (AI) capabilities that these technologies report to offer, resulting in improved accuracy in detecting melanoma compared with standard care.
The intended place in therapy can vary across technologies but is predominantly proposed to be in primary care to assist GPs in triaging suspected melanoma for referral to secondary care.
The main points from the evidence summarised in this briefing are from 13 full-text studies and 1 poster abstract (2 UK studies, 7 German studies, 3 studies from the Netherlands, 1 international study and 1 Canadian study). The 14 studies include over 2,598 people and 570 images from datasets. Most of the studies were done in the secondary care setting. Several studies show that the signal processing and AI in these digital technologies outperformed or were equivalent to standard dermatoscope examination in identifying melanoma.
Key uncertainties around the evidence or technologies are that there is a lack of randomised controlled trials and prospective cohort studies to assess any of the tools in the intended use, clinical setting and population. There are no studies identifying the impact of these tools on clinical decision making by GPs. Only 2 companies (nomela and DERM) have UK‑based studies. Out of the 14 studies, 12 were not based in the UK, which may limit their generalisability to the NHS setting.
Experts agreed that these are innovative technologies but highlighted significant gaps in the evidence base and uncertainties on the potential system benefits and impact on the clinical pathway. Experts noted that these technologies have potential resource consequences but highlighted that this could vary depending on where the technologies are used in the clinical pathway.
There are general safety considerations around the use of these technologies, which include the risk of false positives leading to anxiety for the person and potentially unnecessary procedures. False negatives could potentially delay diagnosis. The technologies have not been tested on rare skin cancers and there may be uncertainty with using these technologies on people with black or brown skin.
The unit costs of the technologies in this briefing vary in their payment plans from £0.60 per eligible user, £30 to £50 per test use (excluding VAT) and up to £53,574 (excluding VAT) for a single technology. These values are not comparable across technologies without considering the place in the clinical pathway, implementation and staffing costs, as well as additional maintenance costs. All technologies may have the potential to reduce some secondary care consultation appointments, which cost £121 per attendance (excluding VAT).

The British Association of Dermatologists, a professional membership body for UK dermatologists, has produced a position statement on AI interventions highlighting that AI technologies have the potential to improve clinical care and optimise processes through using clinical data to inform best practice and outcomes.