Guide

Are Mole Check Apps Accurate? What the Evidence Shows and How to Use Them Safely

Not reliably. A 2020 BMJ systematic review concluded that the algorithm-based smartphone apps it examined cannot be relied on to detect all cases of melanoma or other skin cancers [1]. The review searched studies up to April 2019, so newer app versions are not covered. Use an app to decide what to show a doctor, never as the reason to skip one.

Woman in her forties on a park bench looking at her phone with a thoughtful, sceptical expression

What the 2020 BMJ review found

Freeman and colleagues reviewed the published tests of skin cancer apps. They found nine studies of six identifiable apps: six studies checked results against a biopsy or follow-up (725 lesions) and three against an expert's recommendation (407 lesions). The authors called the studies small and of poor methodological quality, with selective recruitment and a high rate of images the app could not evaluate [1].

Two results stand out. SkinVision, evaluated in two studies of 252 lesions (61 malignant or premalignant), reached a sensitivity of 80% (95% confidence interval 63% to 92%) and a specificity of 78% (67% to 87%) for malignant or premalignant lesions. In plain terms, in that sample the app flagged about 8 in 10 lesions that were cancerous or precancerous and about 8 in 10 that were not. Given so few lesions, the confidence intervals mean the true sensitivity could be anywhere from about 6 to 9 in 10. Its accuracy checked against expert recommendations was poor [1]. SkinScan, tested in a single study of 15 lesions with five melanomas, found none of the melanomas [1].

The authors added that real-world performance is likely to be poorer than reported, because clinicians, not app users, chose the lesions and took the images [1].

CheckMole photo check · Made by us

Check this mole with a photo

Get a plain-language description of its shape, border and colour, and a suggested next step, in about a minute.

$4.99 per check · one-time, no subscription

CheckMole doesn't diagnose.

An earlier test of four apps

A 2013 study in JAMA Dermatology tested four apps on photos of 60 melanomas and 128 benign lesions, each confirmed by a pathologist. Sensitivity ranged from 6.8% to 98.1%, and specificity from 30.4% to 93.7%. Three of the four apps classified 30% or more of the melanomas as unconcerning. The highest sensitivity came from an app that sent the photo to a dermatologist; the weakest came from apps that used automated analysis alone [2].

Those apps are old, and software has changed. The point that still applies is that results differed widely from app to app, so one app's claim tells you little about another.

What regulation labels do and do not tell you

A regulatory mark means a maker followed a process, not that an app catches your melanoma. The BMJ review noted two CE-marked apps and concluded that the CE-marking process for algorithm-based apps did not give the public adequate protection [1]. SkinVision's own site calls the app a regulated medical device and shows a CE mark, but the page we read gives no sensitivity or specificity figures [4].

In the US, the Federal Trade Commission charged the marketers of two melanoma apps, MelApp and Mole Detective, in 2015 with deceptively claiming, without adequate scientific evidence, that the apps could assess melanoma risk. Two of the three marketers settled (one settlement was still proposed, pending public comment), and the FTC said it would pursue the third in court. These were allegations, not findings [3].

AI used inside the NHS looks different. According to Skin Analytics, its DERM, a Class III CE-marked device, is mainly used after a GP's urgent suspected skin cancer referral, in an imaging clinic before the dermatologist appointment [8]. A clinic runs that setting, not a person photographing a spot at home. Ask any app for its certificate and the study behind it.

What a photo cannot show

Dermoscopy is examination of the skin with skin-surface microscopy, which shows structures and patterns that are hard to see with the eye [5]. A phone photo is not a dermoscopy image. The BMJ review also found a high rate of images that apps could not evaluate [1], so blur, shadow and glare matter.

A photo also cannot replace the rest of an examination. A clinician can feel whether a lesion is raised or firm, look at all your other moles, and weigh your history. DermNet notes that AI is not going to replace medical experts in the near future, that a person has to choose which lesion to assess, and that AI can offer a second opinion that can screen out an entirely benign lesion [7].

Change over time is the other gap. A change in a mole's size, shape or colour is a reason to see a GP [9]. One photo cannot show change; two photos taken the same way can.

Safe-use rules

These rules apply to any app, including ours:

- Treat the result as a prompt to act, not a verdict. A reassuring result does not cancel a red flag. - See a doctor for a mole that has changed, itches, hurts, bleeds, crusts or will not heal, whatever the app says [9]. - Take photos the way dermatologists ask: natural, bright light, a coin or ruler for scale, one wide shot, one with nearby skin, one close-up, sharp and without filters [6]. - Take the same photo again after a few weeks, in the same light and from the same distance, so change shows. - Check what the maker claims. An app that says it diagnoses or rules out cancer is making a stronger claim than the evidence above supports. - Keep booking routine skin checks. An app does not replace them.

How the main options differ

SkinVision rates a spot's risk with AI. Miiskin is designed to track moles over time by comparing photos. Other tools describe or flag what is visible. They do different jobs, so read each one's own description before relying on it. Our guides on the best mole check apps and on SkinVision alternatives compare them in more detail.

CheckMole has a free ABCDE questionnaire, which uses no AI: you answer questions about the spot and see which warning signs it matches. Its AI photo check is paid, $4.99 per check and one-time, and describes what is visible in your photo, such as shape, border and colour, then suggests a next step. It does not diagnose, and we make no claim about how accurate it is. It sits next to a doctor's visit, not in place of one.

When to see a doctor

See a doctor today: bleeding that will not stop with firm pressure, or a wound that is red, hot and spreading, with pus or fever.

Book within 2 weeks: a mole that has changed in size, shape or colour, that bleeds or crusts without an injury, itches or hurts, or a sore that has not healed after about four weeks; a new mole in an adult. Book even if an app said low risk [9].

Book in the next few weeks: a new or unusual mark that has not gone away after a few weeks and shows none of the above.

Mention it at your next check-up: a stable mole you have always had, and you are using an app only to keep photos.

Use a tool to prepare, then book a clinician for anything that has changed. The free ABCDE checker is a good place to note what to mention at the visit.

Go through the free ABCDE questions

Frequently asked questions

Are mole check apps accurate?

Not reliably. A 2020 BMJ systematic review of nine studies found that the algorithm-based apps it examined cannot be relied on to detect all cases of melanoma or other skin cancers, and that real-world accuracy is likely poorer than reported [1].

How accurate is SkinVision?

In the BMJ review, two studies of 252 lesions found a sensitivity of 80% and a specificity of 78% for malignant or premalignant lesions, with wide uncertainty, and poor accuracy against expert recommendations [1]. The review searched studies up to April 2019, so it does not cover newer app versions.

Is a CE mark proof that a skin cancer app works?

No. The BMJ authors concluded that the CE-marking process for algorithm-based apps did not provide adequate protection to the public [1]. Ask the maker for the study behind the claim.

Can an app replace a dermatologist?

No. A phone photo is not a dermoscopy image, cannot show a lesion's feel or your other moles, and a single photo cannot show change over time [5][7]. Use an app to decide what to show a doctor, not whether to go.

Is CheckMole's photo check accurate?

We make no accuracy claim. It is a paid AI description of what is visible in your photo, $4.99 per check, one-time, and it does not diagnose. The free ABCDE questionnaire uses no AI. Neither replaces a doctor.

Sources

  1. Freeman K, Dinnes J, Chuchu N, et al.. Algorithm based smartphone apps to assess risk of skin cancer in adults: systematic review of diagnostic accuracy studies. BMJ, 2020.
  2. Wolf JA, Moreau JF, Akilov O, et al.. Diagnostic inaccuracy of smartphone applications for melanoma detection. JAMA Dermatology, 2013.
  3. Federal Trade Commission. FTC cracks down on marketers of "melanoma detection" apps. ftc.gov, 2015.
  4. SkinVision. SkinVision (company home page). skinvision.com, accessed Nov 2026.
  5. DermNet. Dermoscopy (dermatoscopy). dermnetnz.org, accessed Oct 2026.
  6. American Academy of Dermatology. How to take pictures of your skin for your dermatologist. aad.org, accessed Oct 2026.
  7. DermNet. Artificial intelligence in dermatology. dermnetnz.org, accessed Oct 2026.
  8. Skin Analytics. Frequently asked questions (DERM). skin-analytics.com, accessed Nov 2026.
  9. NHS. Melanoma skin cancer: symptoms. nhs.uk, accessed Oct 2026.

Each source was opened and checked against the sentence it supports. How we write our guides