Mr. Onur Gilleard
Qualifications & Experience
Mr Onur Gilleard is a distinguished consultant plastic surgeon on the GMC specialist register. He previously held an NHS consultant post at St Bartholomew’s Hospital in London, specialising in skin cancer and microsurgical reconstruction. In 2021, Mr Gilleard transitioned from the NHS to dedicate himself fully to private practice, allowing him to focus exclusively on providing personalised, high-quality care to his patients. He is recognised as an expert in laser treatments, having conducted research and developed advanced treatment protocols using cutting-edge laser technology to optimise both clinical outcomes and cosmetic results.
Get started today
Book a Consultation
The problem with mole-check apps is not that they are useless. It is the direction in which they fail. An app that raises a false alarm costs you an appointment you did not need. An app that offers false reassurance costs you months — and in melanoma, months are measured as millimetres of Breslow thickness, which is the single strongest determinant of how the disease is treated and how it turns out.
What the Evidence Actually Shows
This is not a matter of professional opinion. A systematic review of diagnostic accuracy studies published in the BMJ examined algorithm-based smartphone applications for assessing skin cancer risk in adults and found sensitivity that was poor, inconsistent, and below what would be required to safely rule out melanoma (Freeman et al., 2020). Several apps assessed in that period have since been withdrawn or reclassified.
Sensitivity is the relevant measure here, and it is worth being precise about why. A test with poor sensitivity misses cases. In a screening context that is the failure that matters, because the person receiving a reassuring result stops looking.
“The problem is not that apps are useless, it is the direction of the error. A false alarm costs an appointment. False reassurance costs months, and months change Breslow thickness.”— Mr Onur Gilleard, Consultant Plastic Surgeon
Four Things a Phone Structurally Cannot Do
These are not limitations that better software fixes. They are consequences of what a phone camera is.
- 1. It cannot see beneath the surface
- Most of what identifies a melanoma sits below the stratum corneum: the pigment network, irregular streaks, blue-white veil, regression structures, polymorphous vessels. A dermatoscope reveals those by eliminating surface reflection, either with an immersion fluid or cross-polarised filters. A phone photographs the reflection. The diagnostic information is simply not in the image — see digital dermoscopy explained.
- 2. It cannot feel the lesion
- Firmness, thickness and whether a lesion is fixed to underlying tissue are not visible in any photograph. This matters most for the melanoma that is most dangerous: nodular melanoma is often symmetrical, evenly coloured and unremarkable to look at, and is identified by being firm and growing. To a camera it reads as flat and benign.
- 3. It has no baseline and no scale
- Melanoma is diagnosed by change more often than by appearance. Without a controlled previous image there is nothing to compare against, and casual photographs taken at different distances under different lighting are not comparable. Colour is also distorted by ambient light and by the phone’s own processing, and colour is diagnostic information.
- 4. It sees one lesion, not one person
- The single most useful sign in self-assessment is the ugly duckling — the lesion that looks unlike the rest of your moles. That is a comparison across a whole person. An app assessing one photograph in isolation cannot make it.
“An app or a photograph can tell you that a mole looks unusual today, but it cannot tell you that it looked different six months ago, cannot feel that it has become firm, and cannot see the structures under the surface that decide whether it comes off.”— Mr Onur Gilleard
Clinical AI and Consumer AI Share a Word, Not a Method
Apps and clinical systems both describe themselves as AI-assisted, and the resemblance ends there.
| Consumer app | Clinical system | |
|---|---|---|
| Image source | Phone camera, uncontrolled lighting and distance | Standardised imaging plus dermoscopy at fixed magnification |
| Sees beneath the surface | No | Yes — that is the point of the dermatoscope |
| Baseline for comparison | None | Previous visits, captured in identical conditions |
| What it produces | A risk score, delivered to the patient | A flag, delivered to a clinician |
| Who decides | Nobody — or the patient | A consultant, who examines the lesion |
| Accountability | A disclaimer in the terms of use | A named, GMC-registered doctor |
Worth noting: the clinical software makes the same point about itself. The FotoFinder analysis screen prints a line across its own AI output — “Results are based on statistics. Diagnosis is physician’s responsibility.” The manufacturer of a medical-grade system is more cautious about its algorithm than most consumer apps are about theirs. More on that division in how the FotoFinder system works.
Where an App Is Genuinely Useful
Dismissing them entirely would be as unhelpful as trusting them. There is one role they perform well:
As a prompt. If an app tells you to get something looked at and you then get it looked at, it has done something valuable. Anything that moves a person from ignoring a lesion to booking an appointment has earned its place.
The failure mode is the reverse. A low-risk score on a lesion nobody has examined supplies a reason not to attend, and it is the reason people delay. Treat a reassuring app result as carrying no information at all — not as a negative result, but as no result.
If you want to self-monitor properly, photography done consistently is more useful than any algorithm: same distance, same lighting, a ruler in frame, repeated monthly. Method in how to photograph a mole for a clinician and checking your own skin.
A separate question is what happens when a clinician rather than an algorithm reviews that photograph. It is a different service with a different failure mode, set out in where teledermatology falls short.
The Regulatory Position
An application that claims to diagnose or assess risk of disease is, in principle, a medical device and falls under medical device regulation in the UK. In practice the category has been uneven: some apps hold appropriate certification, some describe themselves as “wellness” or “educational” tools to sit outside the regime, and the market has changed considerably since the systematic review above was published.
Two practical checks before relying on anything:
- Does it claim to be a medical device, and can you find the certification? A wellness disclaimer is a legal statement that it is not making a clinical claim — which means you should not treat its output as one.
- What does it say when it is unsure? A well-designed tool escalates. A poorly designed one reassures.
What to Do Instead
- Use the ugly duckling test, not an algorithm. Look for the lesion that does not belong among your others. It requires no equipment and outperforms ABCDE for self-assessment.
- Photograph anything you are watching, properly. Consistent conditions, a scale in frame, monthly.
- Book if it has changed. Watching a stable lesion is reasonable; watching a changing one is the failure that costs. The features are in when to see a doctor about a mole.
- Have it examined with dermoscopy, by someone who can also remove it if that is the answer.
| Item | Price |
|---|---|
| Consultation and mole check | £100 |
| Mole mapping — complete package | £300 |
| Mole removal | £350 |
| Histological analysis | £180 |
Appointments at 101 Harley Street: mole mapping or contact the clinic.
Frequently Asked Questions
- Are mole-check apps accurate?
- A systematic review in the BMJ found sensitivity that was poor, inconsistent and below what would be needed to safely rule out melanoma (Freeman et al., 2020), and several apps have since been withdrawn or reclassified. The concern is specifically about missed cases rather than false alarms.
- Can I trust a reassuring result from an app?
- No — treat it as no result rather than a negative one. False reassurance is the failure mode that causes harm, because it removes the reason to attend. If a lesion is changing, the app’s opinion does not alter what should happen next.
- Why can’t a phone photograph do what a dermatoscope does?
- Because the diagnostic features sit beneath the skin surface. A dermatoscope eliminates surface reflection so light penetrates about a millimetre and returns; a phone records the reflection. The pigment network, blue-white veil and vessel patterns are simply not present in an ordinary photograph.
- Which melanoma do apps miss most?
- Nodular melanoma is the concern. It is often symmetrical, evenly coloured and visually unremarkable, and it is identified by being firm and growing quickly. Firmness cannot be photographed, so the most aggressive subtype is the one least visible to a camera.
- But clinics use AI too. What is the difference?
- The clinical system compares standardised dermoscopic images against previous visits and flags differences for a consultant to assess. The app scores one uncontrolled photograph and returns the result to the patient. The manufacturer of the clinical system prints its own disclaimer on screen: results are statistical, and diagnosis is the physician’s responsibility.
- Is there any good use for them?
- Yes — as a prompt. If an app persuades someone to get a lesion examined, it has done something useful. The problem is only ever the reassuring result.
- Are these apps regulated?
- Software claiming to assess disease risk is in principle a medical device, but the category has been uneven, with some apps positioning themselves as wellness or educational tools to sit outside the regime. If an app carries a wellness disclaimer, that is a statement that it is not making a clinical claim.
- What should I use instead?
- The ugly duckling test — look for the lesion unlike your others — combined with monthly photographs taken in consistent conditions with a scale in frame. Then have anything that has changed examined with dermoscopy.
Share this Post
