The valuable thing a skin cancer algorithm can say is sometimes: this patient can leave the urgent pathway. At Chelsea and Westminster Hospital, Skin Analytics’ DERM technology can identify eligible benign lesions without a dermatologist reviewing the images. Patients can be discharged on the day they receive their results. A scarce specialist appointment becomes available for somebody else.
That is the company’s proposition in miniature. Skin Analytics sells a way to organise care around an AI assessment. Its software examines suspicious skin lesions; its surrounding service helps determine what happens next. Cancer concerns move toward clinicians. Selected harmless lesions move out of the cancer queue. The unglamorous prize is a better appointment book.
The London-based business has been working in NHS pathways since 2020. In its June 2026 launch announcement for a new product, DERM Zero, it reported more than 230,000 patients assessed and over 20,000 cancers detected through the underlying DERM platform. Those are cumulative company figures for established pathways, rather than results from the newly announced smartphone service.
01 / THE FALSE STARTThe Christmas that broke the prototype
Founder Neil Daly grew up in Australia, where sun protection and skin cancer checks were part of everyday awareness. Cancer also affected people close to him. A background in physics, mathematics, mobile innovation and consulting brought him to an appealing question: could computers make reliable skin assessment more accessible? He founded Skin Analytics in 2012.
The early answer was awkward. In an interview with L’Atelier BNP Paribas, republished by the company, Daly described asking dermatologists to explain what made a lesion suspicious. Uneven borders? Different colours? Their judgement resisted neat translation. “I can’t codify that,” he recalled. The team nevertheless built roughly 200 mathematical measurements and combined them using a support vector machine.
Testing looked promising. Then a colleague tried the prototype on relatives at Christmas, and it falsely suggested the whole family had melanoma. This was an informal test, according to Daly’s account, but it exposed the gap between development data and ordinary photographs. Deep learning subsequently improved performance, he said. The team also chose training data tied to tissue-test outcomes, accepting a smaller dataset for firmer labels.
02 / THE USEFUL DECISIONAn exit for the benign cases
DERM stands for Deep Ensemble for the Recognition of Malignancy. In the established NHS workflow described by NICE, a smartphone with a dermoscopic lens captures detailed images. The software assesses them for categories including melanoma, squamous cell carcinoma, basal cell carcinoma, precancerous changes and benign lesions. It supports a decision about further care.
A benign result can lead to discharge with advice about what to do if concerns persist or the lesion changes. A suspected cancer or precancer result goes to an NHS dermatologist for virtual review and a management plan. The product therefore depends on a working connection between photography, image assessment, communication and treatment.
Chelsea and Westminster’s March 2025 announcement put the ambition in operational terms: autonomous triage aimed to free more than 30% of suspected skin cancer appointments. Its 2024/25 annual report subsequently described freeing more than 35% of specialist appointments. The difference between identifying a harmless lesion and releasing an appointment is the implementation work in between.
Illustrative workflow. Eligibility rules and local safeguards apply.
Patients encounter Skin Analytics through several doors. NHS teams commission AI pathways. Bupa’s remote service sends members a phone and magnifying lens, with images and medical history assessed by a dermatologist. Freedom Health offers consultant-led remote lesion and general dermatology services, subject to policy cover. Skin Analytics also advertises UK mole checks and general dermatology through direct access.
These distinctions matter when someone asks what they can do with the company’s technology. An insurer’s dermatologist-reviewed service and an autonomous NHS assessment are different care arrangements. Access, payment, equipment and the person responsible for reviewing the result depend on the particular service.
03 / READ THE DENOMINATORTwo percentages, two different questions
Skin Analytics’ performance page reports a melanoma negative predictive value of 99.9% and melanoma sensitivity of 97% for December 2023 to November 2025. The summary covers 90,236 lesion outcomes overall, with separate denominators for individual measures. These figures deserve more attention than a catch-all claim of “accuracy.”
Sensitivity asks how many actual melanomas the system identifies. Negative predictive value asks how often a negative result is correct. The latter depends partly on how common melanoma is in the population being tested. A very high negative predictive value does not mean that the system finds the same proportion of all cancers.
Of actual melanomas, how many were identified?
Of negative results, how many were correct?
NICE’s recommendation is equally specific. Its early value assessment allows DERM in NHS teledermatology for adults referred through the urgent suspected skin cancer pathway while more evidence is collected. Safety measures include a healthcare professional checking assessments of lesions in black or brown skin. Regulatory certification and a recommendation to collect more evidence answer different questions.
There are also cases outside the tool’s appropriate use. The company identifies ulcerated lesions and lesions under nails as examples routed to a dermatologist. A service that cannot handle unsuitable images, excluded lesions or an unresolved patient concern has an unfinished clinical process, however impressive its average performance looks.
04 / FOLLOW THE APPOINTMENTSThe £30 question
NICE’s May 2025 technology description lists £30 per referral for the online DERM assessment. A virtual review by NHS teledermatology staff adds £8.20. It also describes a combined £35.90 price when subsequent biopsy results are shared with the company. These are published reference prices for that evaluation, rather than a universal tariff for every contract or newer product.
Earlier NICE documentation describes pay-per-use and block contracts. The commercial logic is straightforward: healthcare organisations pay for assessment and associated services, while insurers offer a route to members. The economic argument depends on the work avoided downstream. A software fee can be worthwhile if it removes enough appointments, reviews or procedures.
It can also become an extra expense. An independent assessment of company economic models highlighted how results depended on discharge rates and assumptions about ordinary teledermatology. Keeping a second review can consume the very specialist time automation is supposed to release. Buyers need to count the entire referral journey, including cases the AI cannot assess.
“Diagnostic accuracy is table stakes”
Neil Daly · Maddyness interview, May 2022
The company’s financing provides another measure of the undertaking. A £4 million Series A was reported in September 2020. In April 2025, Skin Analytics announced £15 million in Series B funding led by Intrepid Growth Partners. The stated plans included broader dermatology products and expansion into Europe, Australia and the United States. Fundraising finances those ambitions; it does not demonstrate that each market is already open.
05 / THE WORK AROUND THE MODELA medical device needs more than engineers
Skin Analytics’ management spans computer vision, practising medicine, clinical research, technology and dedicated quality and regulatory affairs. That mix is instructive. A hospital is buying a system that must keep working when a photograph is poor, a patient is unsuitable or a clinician questions the result. Publishing software is one part of that obligation.
The company’s public account of its culture centres on improving access and patient outcomes. Its organisational choices give that language substance: regulatory and clinical functions sit alongside research and engineering. DERM does not continuously teach itself on incoming cases. Updates are manually released after validation and the appropriate regulatory process, according to its FAQ.
Its competitors include the existing service: a dermatologist reviewing photographs remotely or seeing patients in person. NICE has also examined products such as Moleanalyzer pro and SkinVision, whose intended uses and delivery models differ. Skin Analytics’ distinctive position combines regulated autonomous decisions with the practical machinery of NHS referrals.
For founders and hospital managers, the useful lesson is to test whether a product changes the next action. Agree which patients qualify, how exceptions move to clinicians and which outcomes will be checked. Measure specialist work actually released. The Christmas prototype is a memorable reason to demand prospective evidence from the setting where a product will be used.
06 / THE NEXT CAMERANow remove the special lens
DERM Zero, unveiled at HLTH Europe in Amsterdam on June 16, 2026, takes aim at another obstacle: equipment. Skin Analytics says its Class III regulated technology can now deliver autonomous assessment using a standard smartphone without a dermoscope. The intended settings include primary care, pharmacies and community services, delivered with healthcare partners.
The product page describes taking a location photograph and a close-up, followed by analysis and an outcome linked to the provider’s care pathway. That changes where an assessment might begin. It also makes the route from a concerning result to a real appointment essential. Easy image capture is useful only when the resulting decision leads somewhere.
European distribution relationships were already taking shape. The company announced an Affidea partnership in August 2025 and a Nordic partnership with Human Bytes that November. The move abroad will require local delivery arrangements as well as technology. Its own website continues to describe DERM as investigational in the United States.
Skin Analytics’ ambition is a future without deaths from skin cancer. Its immediate proposition is more measurable: help the right patient reach the right clinician, and let others leave an urgent queue with appropriate reassurance. The next test is whether that service can travel from hospital imaging rooms to the phones people already carry.