Rise of the robot dermatologist: Could Google AI really help patients identify skin conditions?

Big tech says artificial intelligence could soon help patients make better-informed decisions about their health – but, asks Kate Ng, how well can it really work?

Google’s new AI tool aims to help users get more information on potential skin, hair or nail conditions (Getty Images/iStockphoto)

The future is here but instead of hoverboards and holidays to Mars, technology might just be able to help you figure out just how seriously you should take that mole on your arm.

Google has announced it plans to launch a tool powered by artificial intelligence (AI) that can help users identify skin, hair and nail conditions just using their smartphone camera. The tech giant’s new “dermatology assist tool”, which has taken three years to develop, is currently being trialled with the brand hoping to launch the pilot later this year.

Based on tools previously developed by Google to improve the screening process for certain cancers and tuberculosis, the web-based dermatology-assist application uses some of the same techniques.

With nearly 10 billion Google searches related to skin, nail and hair issues every year, the AI tool could be a useful way to either allay fears or spur a person into action to get their condition assessed by a doctor, claims the brand.

“While many people’s first step involves going to a Google search bar, it can be difficult to describe what you’re seeing on your skin through words alone,” writes Peggy Bui, a product manager at Google Health.

Instead of simply typing into the search bar, users can find out more by taking photos of the area of concern and answering a series of questions about skin type and symptoms to help the tool narrow down the possibilities.

Recognising and correctly identifying a vast array of skin conditions is not Google’s only challenge if this tool is to be successful

Google’s AI can then identify up to 288 conditions and will give users a list of possible matches. For each, it will also show users information and answers to commonly-asked questions, as well as sharing images from the internet to help users determine their skin condition further.

However, the company warns the tool “is not intended to provide a diagnosis nor be a substitute for medical advice”. Instead, it aims to give users “access to authoritative information so you can make a more informed decision about your next step”, Bui says.

While it has been CE marked – showing the manufacturer has checked that these products meet EU safety requirements – as a Class 1 medical device in the European Union, it has not yet been evaluated by the USA’s Food and Drug Administration (FDA) for safety or efficacy.

Of course, recognising and correctly identifying a vast array of skin conditions is not Google’s only challenge if this tool is to be successful. In recent years, engineers and computer scientists have raised the issue of race and gender bias within algorithms, rendering such systems incapable of recognising darker skin tones or women due to a lack of diversity in development.

The issue, dubbed the “coded gaze” by Ghanaian-American computer scientist Joy Buolamwini, forced those developing AI algorithms to examine their existing biases and create more diverse systems.

Google promises their dermatology assistant tool “accounts for factors like age, sex, race and skin types – from pale skin that does not tan to brown skin that rarely burns”. The company said the model has been “fine-tuned” and has been fed around 65,000 images and case data of diagnosed skin conditions, as well as millions of images of skin concerns and healthy skin examples encompassing different demographics.

Dr Tanya Bleiker, the president of the British Association of Dermatologists, told The Independent that while AI’s potential in the dermatological field is “very exciting” and much research is being done to see if it can support the work of professionals and be put in the hands of consumers, there is a need to continually ensure the technology is up to scratch.

The safety of dermatology patients and the wider public must be the number one priority

“The safety of dermatology patients and the wider public must be the number one priority. Despite this rapidly advancing field and the commercial drive to adopt AI into clinical pathways as soon as possible, at the present time the evidence-base is weak, and more research is needed to make sure the technology is safe for patients and that it will improve their care,” Dr Bleiker says.

“No shortcuts can be taken, either in the research, testing, or regulatory processes. After all, extraordinary claims require extraordinary evidence.”

She also says there is the wider problem of addressing a pressing shortage of consultant dermatologists in the field dealing with increasing rates of skin cancer, the UK’s most common cancer.

She emphasised that investing in the workforce and better training for healthcare professionals is the key to solving this problem. “We should also be mindful not to let the future potential of AI, and other technological solutions, prevent us from acting now to solve the workforce shortages in dermatology and improve training in dermatology for all healthcare professionals,” Dr Bleiker says.

As science continues to offer novel ideas and solutions for the future of healthcare, we must ensure the technology delivers the level of service needed across the board, and the focus on innovation does not distract from the real issues at hand.