← Back to Home

Southeast England expands AI skin cancer triage, allowing benign lesions to bypass specialist appointments

DERM will be rolled out to 18 NHS trusts across the region, allowing some patients with suspected skin cancer to leave the cancer referral pathway earlier. However, projected savings in clinical time remain model-based estimates, and the safety of autonomous assessment requires continued monitoring.

By SURL BioNews

A suspicious skin lesion can leave patients facing weeks of anxiety while waiting for a specialist appointment. Southeast England is expanding AI triage in the hope of addressing benign lesions earlier and reserving dermatologists’ time for people who need further investigation. The key change is that some patients can leave the suspected cancer referral pathway without an individual consultation with a dermatologist.

Skin Analytics announced on October 6, 2026, that its DERM system would be rolled out from the autumn to all 18 NHS trusts in Southeast England, extending services to cover approximately 9 million people. The company described this as the largest deployment of autonomous medical AI in NHS history; that characterization comes from the company and does not mean every trust has already gone live.

DERM processes images of lesions already referred for suspected skin cancer. After healthcare staff capture images using a dermatoscope, the system determines whether specialist assessment is needed. Eligible cases assessed as benign can leave the cancer referral pathway at the triage stage, while higher-risk cases proceed to specialist examination. This differs from population-wide skin cancer screening, and an image-based triage result cannot be treated as a definitive pathological diagnosis.

The benefits of the three-year rollout plan currently come primarily from model-based estimates. According to the company’s announcement, up to approximately one-third of urgent suspected skin cancer referrals could be triaged and discharged from the pathway before the first outpatient appointment, potentially freeing up an estimated 13,500 hours of specialist time annually. These figures are not yet measured outcomes from this regional deployment. Whether they translate into shorter waits for treatment will still depend on capacity for subsequent investigations and treatment.

Existing evidence provides some support, but requires careful interpretation. A white paper prepared by Edge Health in 2024 for NHS England’s Outpatient Recovery Programme analyzed 33,693 lesions across 10 care pathways and found that DERM had a negative predictive value of 99.8% for ruling out melanoma. This means that approximately 99.8% of lesions classified as non-melanoma were indeed not melanoma; it does not mean that 99.8% of cancers were detected, nor that all types of skin cancer were ruled out.

Autonomous use remains subject to conditions. NICE’s early value assessment allows DERM to be used in NHS teledermatology services while evidence is collected to continue establishing its clinical effectiveness and cost-effectiveness. Its implementation recommendations also state that patients with darker skin tones classified as Fitzpatrick types 5 and 6 should receive a second assessment by a healthcare professional. In other words, expanding deployment has not eliminated gaps in the evidence for different skin tone groups.

This rollout brings AI into care decisions with more direct consequences: determining who can leave the suspected cancer pathway earlier. Assessing its impact will require tracking not only how many outpatient appointments are saved, but also whether triaged patients experience missed or delayed diagnoses, and whether those who need treatment receive care sooner. These outcomes will be needed to establish whether wider deployment improves patients’ circumstances.

References

  1. Lelezard
  2. Skin Analytics