Why we write on this topic:
Skin cancer is the most common cancer in the Netherlands: about half of all cancer diagnoses concern skin cancer diagnoses. A report by the Netherlands Comprehensive Cancer Organisation (IKNL) indicates that these numbers are set to increase rapidly over the coming years. Within Europe, the Netherlands is, second only to Norway, even the country with the highest incidence of skin cancer.
Back in 2019, Kirill Sokol founded Skinive: a start-up that can detect skin diseases at an early stage with the help of a smartphone app. By taking a photo of a suspicious patch of skin with your smartphone, artificial intelligence (AI) can be used to determine whether your skin has any signs of skin disease. In other words, it is a kind of digital self-test. More than 1,000 dermatologists and 60,000 patients in Europe are now using the app. Nevertheless, some experts are critical.
Sokol explains how the app works, “It basically works the same way as Google Images. You take a picture, put it in a database and then it is cross-referenced with other images of skin disease. This is one way of determining if you are at risk of developing a skin disease such as cancer.” Linda Summer of the KWF Dutch Cancer Society calls Skinive’s app a “fantastic development,” provided it does not lead to cases of overdiagnosis. “Therefore, if people see a patch that they don’t trust, we always recommend that they contact their doctor.”
A self-test every week
Sokol emphasizes that the “self-test” does not provide an official diagnosis, but it does give you a clear picture of a potential diagnosis. Skinive’s database has access to more than 650,000 images that have been sourced from medical professionals and app users. The more pictures users upload, the more reliable the comparisons and thus the advisories become. The company recommends taking a self-test every week. According to dermatologist Soe Janssens of the Dutch Antoni van Leeuwenhoek Hospital (AVL), this is “utterly over the top”, as, “when someone has had melanoma, we check the entire skin twice a year at most. Getting someone who hasn’t had skin cancer to do their own skin inspection every week is unnecessary.”
Too simple to upload a blurry photo
The dermatologist tested out the app for this article. She Is critical about it: “I noticed a bunch of limitations. For example, it does not take into account a patient’s story, something that is really important in order to make a good assessment of a skin patch.” She also raises questions about the use of phone cameras: “If you upload a blurry photo, you don’t get any kind of notification that the photo is too blurry. That’s crazy, because you can’t get proper advice that way, can you?” Several studies, including a study published in the Netherlands Journal of Medicine, show that similar apps have a low positive predictive values. The study found that a large proportion of app users receive false warnings about possible skin cancer and are quickly advised to seek medical attention.
According to chief medical technology scientist Arjen Amelink of the Vrije Universiteit of Amsterdam, this is easy to clarify. “Most likely, they want to be on the safe side as a company. Of course, you don’t want to say that there is nothing wrong if it does turn out that someone has cancer. I wonder how they safeguard the boundary between sending someone through to get help and saying that nothing is wrong.”
According to Sokol, when users use the app, they are actually directed to the right specialist, something that often doesn’t happen at present. “By doing our self-test, you immediately know which specialist to go to – a dermatologist, beautician, allergist, oncologist or an infectiologist. At your GP, you are often referred to a dermatologist without knowing if that is the right specialist to go to. And for an appointment to see a dermatologist, you spend an average of two months on a waiting list. By being sent to the right specialist the first time around, you save time and money.”
Health claims need to become substantiated more
Scientist Amelink sees more and more initiatives like Skinive. Yet he has his reservations about the method. “I wonder to what extent that this artificial intelligence is sufficiently trained to give well-founded advice. Every phone camera works differently in terms of light and quality. That could be a factor in the diagnostics.”
He compares Skinive to Skinvision, a similar smartphone app that has been undergone scientific research. “Most health apps still lack proper laws and regulations. This means that a lot is feasible now, but there is still not much oversight as to whether it actually works.”
According to Amelink, however, that is bound to change in the future. “There’s going to be more regulation on these types of initiatives and then a lot of companies will fold. If an app makes a certain health claim, it has to be scientifically substantiated in order to continue to exist.”