{"id":45364,"date":"2024-02-17T09:38:29","date_gmt":"2024-02-17T08:38:29","guid":{"rendered":"https:\/\/drromanovaleria.ch\/?p=45364"},"modified":"2026-06-09T16:24:29","modified_gmt":"2026-06-09T14:24:29","slug":"artificial-intelligence-age-estimation","status":"publish","type":"post","link":"https:\/\/drromanovaleria.ch\/en\/blog\/intelligence-artificielle-estimation-age\/","title":{"rendered":"Une IA peut conna\u00eetre votre \u00e2ge r\u00e9el \u00e0 partir d\u2019un simple selfie"},"content":{"rendered":"

AN AI CAN KNOW YOUR REAL AGE FROM A SIMPLE SELFIE<\/h1>\n

AI for estimating biological age from a simple selfie<\/h2>\n

Imagine a simple photograph of your face \u2013 a quickly taken selfie \u2013 revealing not only how many years old you are, but also your \"biological age\", meaning your body's actual condition. The idea might sound like science fiction, but it is now embodied in an artificial intelligence system called FaceAge, recently presented in The Lancet Digital Health<\/em>. This technological achievement opens unprecedented avenues in medicine, but also raises ethical, methodological and reliability questions.<\/p>\n

\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"Photo
\n\t\t\t\t\t
\n\t\t\t\t\t\t\t\t\tBook an appointment
\n\t\t\t\t\t<\/a><\/p>\n

\n\t\t\t\tContents\t\t\t<\/h2>\n

\t\t\t\t\t\t\t\t\t\t<\/path><\/svg>
\n\t\t\t\t<\/path><\/svg>
\n\t\t\t\t<\/path><\/svg>\t\t\t<\/p>\n

From chronological age to biological age<\/h2>\n

Chronological age, measured in years, only imperfectly reflects our actual health. Two people of the same age can differ considerably in vitality, resilience or frailty. Biological age seeks to express this disparity: it describes the actual condition of tissues and organs, or each person's individual pace of ageing.
Until now, assessing biological age relied on complex, often invasive indicators: blood tests, telomeres, inflammatory or metabolic markers. FaceAge aims to make this estimate available within seconds from a simple facial image.
The intuition behind this programme is simple: because the face reflects skin structure, tissue density, muscle tone and overall bodily harmony, it acts as a mirror of general health. By analysing thousands of details imperceptible to the human eye, artificial intelligence could extract a biological signature of ageing.<\/p>\n

FaceAge<\/h2>\n

FaceAge emerged from collaborative work involving several hospitals and universities, including Harvard Medical School, Massachusetts General Hospital and the MAASTRO centre in the Netherlands. Researchers designed a deep learning model capable of learning from facial images.
The algorithm was applied to nearly sixty thousand photographs of people aged sixty and over considered healthy. These images came from public databases containing men and women of different ages and backgrounds. The system was then tested on more than six thousand patients with various cancers, whose photographs and clinical data were available.
The principle was to compare facial age estimated by artificial intelligence with actual chronological age, then correlate the difference with survival and disease progression. The results are striking: among cancer patients, facial age calculated by the algorithm exceeded chronological age by an average of nearly five years. The larger this difference, the poorer the survival prognosis.
This correlation remained significant after adjustment for age, sex, tumour type and other clinical variables. In other words, the face revealed accelerated ageing associated with disease severity, and FaceAge detected it more accurately than the human eye.<\/p>\n

How does FaceAge \"see\" your biological age?<\/h2>\n

A face is more than a combination of features: it is a complex topography of volumes, densities, colours and imperceptible movements. When analysing an image, artificial intelligence does more than identify wrinkles or shadows; it studies thousands of points, skin density, contours, light distribution and imperceptible geometric variations.
The FaceAge model is not explicitly programmed to seek particular signs of ageing. It learns independently to identify recurring patterns statistically associated with a given age or biological condition. It thus captures signals invisible to the naked eye, perhaps linked to microcirculation, dermal firmness or subcutaneous tissue composition.
What humans interpret as an overall impression \u2013 a tired face, lively eyes, dull skin \u2013 becomes an infinite number of measurable parameters for the machine. This mathematical precision, free from perceptual bias, allows FaceAge to exceed human assessment capabilities.
Nevertheless, researchers emphasise that the system does not rely on superficial criteria such as hair colour or visible wrinkles. It appears to analyse more structural facial aspects, less influenced by aesthetic enhancements.<\/p>\n

The clinical value of artificial intelligence<\/h2>\n

Researchers designed FaceAge as a medical marker rather than a cosmetic tool. In cancer care, its value lies in the possibility of assessing a patient's physiological frailty without invasive examinations.
Results published in The Lancet Digital Health show that, in oncology, facial age estimated by FaceAge is an independent prognostic indicator. The higher facial age is relative to chronological age, the greater the likelihood of mortality. In some cohorts, incorporating FaceAge into prediction models significantly improves survival estimate accuracy.
Even more surprisingly, when clinicians had to predict the probability of death within six months in patients with advanced disease, their performance improved markedly when given the artificial intelligence result. This suggests FaceAge detects clinical signs invisible even to specialists.<\/p>\n

The face and artificial intelligence<\/h2>\n

These results open a new field: medicine based on automated facial analysis. In the near future, simple photographs incorporated into electronic medical records could conceivably assess a patient's biological condition, monitor treatment progress or anticipate certain risks.
Beyond oncology, the method could extend to other chronic or degenerative diseases. Researchers envisage applying this algorithmic reading to cardiology, geriatrics or preventative medicine. Long observed by doctors for visible signs of fatigue or suffering, the face would become a quantitative, objective and measurable biomarker.
In aesthetic medicine, this approach could also mark a turning point. If facial biological age becomes measurable, it could be possible to assess the real effectiveness of anti-ageing treatments, compare skin changes before and after a procedure, or personalise care according to each patient's biological profile.
For now, however, these applications remain hypothetical: FaceAge is primarily a research tool, and its reliability still needs strengthening.<\/p>\n

The limitations of artificial intelligence in aesthetic medicine<\/h2>\n

The study's authors acknowledge several limitations. The algorithm was trained on predominantly Western databases, with limited representation of darker skin phototypes and non-European facial structures. Yet ethnic differences, lifestyle habits, sun exposure and cultural characteristics can influence signs of ageing.
Furthermore, photographic conditions \u2013 lighting, posture, expression, make-up and camera quality \u2013 can alter estimates. The model therefore needs refinement to withstand these variations and avoid systematic bias.
It will also be necessary to determine how far aesthetic procedures, skincare or medical treatments can change the facial age perceived by artificial intelligence. If a face is artificially altered, is the signal's biological value distorted? These questions remain open.
Finally, the exact mechanism by which the algorithm associates particular facial characteristics with biological age remains unclear. Like most deep learning models, FaceAge acts as a \"black box\" whose internal criteria are not entirely interpretable. For now, this opacity limits its understanding and clinical acceptance.<\/p>\n

Ethical and societal issues surrounding AI age determination<\/h2>\n

A machine's ability to infer medical information from an ordinary image raises major questions. Facial photographs, so commonplace in daily life, could become health markers or even risk indicators. Who would then have the right to access them? How can the confidentiality of biomedical data derived from a simple photograph be guaranteed?
Researchers emphasise the project's exclusively medical and experimental nature. However, in a world of ubiquitous selfies and increasingly accessible artificial intelligence tools, misuse of such technology for non-medical purposes (insurance, recruitment, surveillance) cannot be ruled out.
Ethically, the boundary between knowledge and intrusion becomes blurred. Knowing a person's \"real age\" also means entering the biological privacy of their body. If such information were communicated to a patient, it should be done cautiously within a supervised therapeutic framework.<\/p>\n

How artificial intelligence reads a face<\/h2>\n

FaceAge is part of a broader movement in which the face becomes a vehicle for medical information. Science is rediscovering what clinical medicine intuitively knew: the face is not only a social symbol but also biological evidence. Artificial intelligence restores its role as a universal indicator, freeing it from the subjectivity of human observation.
However, this fascination with measurement and prediction must not obscure a fundamental truth: biological age, even if calculable, does not define a person. It expresses neither their history nor their emotions nor the richness of their vitality. Artificial intelligence can decipher the surface, but it will never fully grasp what makes a human face unique.
FaceAge therefore does not replace the doctor's assessment.<\/p>\n

\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\"Photo\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/p>\n

Article written by Dr Romano Valeria<\/a><\/p>\n

SHARE THIS ARTICLE ON<\/p>\n

\t\t\t\t\t\t\t\t<\/path><\/svg>
\n\t\t\t\t\t\t\t\t\t\tLinkedIn\t\t\t\t\t\t\t\t\t<\/p>\n