Connect with us


Is Google’s new AI dermatology tool taking the correct approach?

Avatar photo



Google has announced that it will be launching a new AI-enabled app later this year that can detect potential dermatological issues. The tool aims to make it easier for the public to understand issues related to their skin, hair and nails, however one expert warn that it could lead to unnecessary panic, Health Tech World reports. 

Google recently unveiled a new tool that uses artificial intelligence to help detect dermatological issues on the skin, hair and nails. 

The web-based application, which has already been awarded a CE mark for use as a Class I medical tool in Europe, is expected to launch as a pilot later this year. 

The AI system can identify 288 different skin conditions however, Google stressed that the app is not a substitute for medical diagnosis. 

It was developed over a three year period with the help of 65,000 images, millions of curated skin concern images and case data of diagnosed skin conditions. 

Users simply take three images of the skin, nail or hair concern with their phone camera then answer a series of questions. The AI tool analyses the information and provides a list of possible matching conditions that the user can choose to research in more detail. 

Google says it sees almost 10 billion Google searches related to skin, nail and hair issues every year. In Google’s view, this tool is more effective than searching for the information on the internet. 

However, this leads to an interesting debate as to whether we should be identifying everyone who potentially has a disease. 

Bruce Murray, Technical director of the British med-tech company Moletest, is of two minds and warns of the potential danger of Google’s dermatology tool. 

Moletest is developing the first and only ‘rule-out’ device for melanoma skin cancer which rules out patients with non-urgent skin lesions and reduces the number of unnecessary dermatology referrals from GPs to secondary care for suspected melanoma. The device is known as Nomela. 

“The danger is that people panic,” Murray told Health Tech World. “And this is why we don’t see Nomela being at the front door of primary care. 

“The nature of these ranges within the reading where the melanoma sits is that under that range are lots of benign [lesions]. 

Latest HT World video interview

“I think it might end up with more worried people going to their doctor because [Google] have had to play it safe. They’ve got to set it up so that it doesn’t mislead people into thinking that something is benign if it isn’t.”

Murray also believes that there is an issue with the quality and consistency of images that are produced by members of the public. 

Murray continued: “It depends on how the photo has been taken; what device, what camera, what pre-processing, what lighting. It’s not simple, which is why we need to train people and certify that they really know how the system works. You don’t normally take a picture of your skin and it’s actually a very difficult thing [to do].

“I think the main difficulty they’re going to face is the variability of the quality of images. You’re looking for things that are so subtle, so that if you have some other variable which is greater than that subtle change, it is going to be very difficult to get the accuracy you need.

“Trying to analyse a skin condition is really difficult based on an image and it’s understandable that dermatologists are very skeptical about the ability of a [machine] to do it reliably.”

Despite having concerns about Google’s new tool, Murray was impressed with the company for developing a tool that can cover such a wide range of conditions. 

“I’m full of admiration for Google that they’ve managed to train a system to 288 different conditions. I think it’s brilliant, what they’ve done and I’m sure it will be widely used by people who are worried. 

“I think it’s much better that they’ve done it rather than allow people to just search aimlessly on the web.”

Moletest originally launched in 2010 to create a diagnostic tool that could identify six conditions, but the team had to “give up” due to a lack of consistent, high-quality data. 

Moletest changed its direction away from diagnostics and decided instead to focus on building a tool called Nomela, that could rule out melanoma rather than identify it. 

Nomela is built into an Apple iPad and is used by a clinician to assist their decision-making process. Like Google’s tool, Nomela analyses images taken from the mobile device and in a matter of minutes can give an instant decision as to whether the patient is at high or low risk of having melanoma. 

“We’re technically trying to do something which is much simpler and less threatening to the patient and the professional,” Murray said. “It is more likely to have an actionable impact and [patients] can be reassured straightaway.

“It’s a safe first step towards the more automated approach to medicine.”

Melanoma is a relatively rare disease but clinicans must be careful not to rule out possible cases early on in the process. This means the majority of patients will progress along the pathway despite not having the condition. 

In a clinical trial at Addenbrooke’s Hospital, Moletest is analysing patients that have had their lesion of concern removed for a biopsy. Of this cohort, only 1 in 10 had melanoma. Data from an earlier clinical trial looking at a cohort of patients that had been referred by their GP, only 3.5% had melanoma. 

According to Moletest, Nomela could help to reduce the number of unnecessary dermatology referrals to secondary care by over 50%, potentially saving the NHS £125M per year. 

Both Nomela and Google’s dermatology assist tool are aiming to tackle a similar problem. In the dermatology field, there is a significant issue surrounding supply and demand. 

The British Association of Dermatology recently revealed that 30% of posts are currently unfilled and as the government has implemented a maximum two-week wait for people who might have cancer, those with less serious conditions are often unable to see a dermatologist. 

“Supply and demand is hugely out of balance [and] dermatologists are overwhelmed by this demand,” Murray said. “If you can have a significant effect on the demand, the queue shortens over a period of time and then [dermatologists] have the time to see people that have serious skin conditions that may be very debilitating and need treatment. 

“You’ve got to see a specialist; you can’t deal with it in primary care. that’s exactly where we positioned ourselves, either on entry into secondary care, or in the GP clinic.” 

Continue Reading
1 Comment

1 Comment

  1. Pingback: AI and computer vision could transform kidney treatment and save NHS millions - Health Tech World

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending stories