Anastasia Georgievskaya is using AI to develop biomarkers from photos of consumers’ skin. She is doing important work for the longevity revolution, but the consumers are more attracted by offers of improved attractiveness than offers of extended lifespan.
Skin biomarkers
Anastasia Georgievskaya runs a company in Estonia that provides consumers with recommendations engines for healthcare and lifestyle changes, using indicators from photographs of their skin. The company is called Haut.AI, and it operates a software platform which solicits images, and also lifestyle and preference data from individuals. It processes this information, and provides the results to clients.
The clients include major skincare brand owners, but also service providers like health clinics, surgeries, and providers of nutritional and general health information and advice. The Haut.AI platform facilitates these clients’ research and development programmes, provides them with a better understanding of their customers’ needs, and enables them to build their own skin analysis apps.
The data on the platform includes over a hundred skin biomarkers, which enables Haut.AI and its clients to build anonymised customer profiles, and provide personalised recommendations. In time, the company intends to launch its own direct-to-consumer platform based on their large data set.
Georgievskaya is Russian by birth, and was educated as a biophysicist in Moscow. During a hackathon in 2014 she got into a conversation about the chronic shortage of data which can indicate a person’s biological age – as opposed to their chronological age. Most of this data is generated by blood or tissue samples which are sent to laboratories for processing. This imposes a delay which can be as much as three months, although this is being reduced – for instance gene sequencing is now often completed in less than a month.
AI’s Big Bang and image recognition
Two years previously, computer researcher Geoff Hinton and his team had ignited what became known as the Big Bang in AI by creating a neural network AI which could recognise images almost as well as a human. Georgievskaya wondered whether this kind of image recognition AI could be applied to skin images, to develop a valuable biomarker. She spent two years investigating what could be achieved, and experimenting with different approaches to capturing and using images of skin.
Photos can capture a range of indicators of aging, including uniformity of colour and texture, sagging, wrinkling, eye bags, spots, and so on. Photos can obviously be taken of skin from various parts of the body. Faces are good indicators of aging, but they have two problems. First, many people apply ointments, oils and unguents to their faces, and some people have plastic surgery, or inject Botox. Second, people are understandably nervous about submitting images of their faces to a software platform, as they seem harder to anonymise.
Other parts of the body are harder to specify, or more personal, or harder to photograph. Hands suffer none of these drawbacks. There is the issue that the skin on our hands is affected by our professions and our lifestyles: a farmer and a beautician will have very different hand skin. But it turns out these differences can be adjusted for.
AI image analysis
Having selected hand skin as the principal target, Geogievskaya had to work out how to train consumers to submit photographs that met acceptable standards. By now, most people had a phone with a camera, but it was hard to teach people how to obtain the right light conditions, focus, and so on. Even people who are smart and willing to follow instructions often don’t manage it in practice, and of course different phones have different types of camera.
The solution was to develop an AI which analyses each submitted photo against 17 metrics, and accepts an image only when it meets all the conditions. Consumers don’t download an app to submit the photos. The images are analysed partly in the web browser and partly in the cloud.
Appearance beats health
When it came to inviting consumers to submit their skin images, Georgievskaya was surprised to find that people are more motivated by offers to improve the appearance and attractiveness of their skin than to improve their health. This is not just vanity: most people are sceptical about offers to extend their healthy lifespan, and less so about offers to improve their appearance. This is presumably a testament to the effectiveness of the cosmetics industry – or at least its marketing departments.
Consumers are particularly keen to look younger than they are, and to be associated with healthy-looking, attractive groups. Plastic surgeons have long pointed out that people feel happier and more comfortable if they believe they look younger. Interestingly, when people feel younger they behave younger, which can create a virtuous spiral if they engage in the activities of younger people, including sport and exercise.
Perhaps you are wondering whether this is more true of men than women. It is true that most cosmetic products are sold to women, and indeed there has long been a stigma about men paying attention to their own skincare. The situation is improving, but there is still a long way to go. Male cosmetics are no longer a small niche market, but many men have yet to consider purchasing skincare products.
This is reflected in the ads directed at men by skincare brands. In 2018, male grooming product manufacturer Old Spice started a campaign (here) with the strapline “Men have skin too”, exhorting men to use products with benefits like moisturization and exfoliation. Earlier this year, Lumin Skincare launched an ad (here) in which one man “mansplains” to another the danger of neglecting skincare, and describes its products as “a workout for your face”.
Democratising anti-aging
Many of Georgievska’s colleagues in the longevity research field are enthusiastic about her work, which they see as democratising access to biomarkers of aging. Only if we can track the impact of therapies and lifestyle changes on people’s biological age – as opposed to their chronological age – can we assess their value, and persuade people to adopt them. But other colleagues are resistant, preferring biomarkers obtained through bloodwork and tissue samples. Georgievskaya is optimistic this will change, not least as the cameras and other sensors available on our phones and our wearables get better over time.
She is also optimistic, albeit cautiously, about the prospects for increasing longevity. She thinks that most people who are 30 today should become healthy centenarians, barring accidents, genetic defects, or infections. She foresees particular opportunities for healthy longevity in women. First-time mothers over 60 years old might become unremarkable, which would be particularly liberating for career-minded young women.
