A photo of the face instead of tests? AI is to assess the biological age and prognosis of cancer patients
Researchers at Mass General Brigham in Boston have developed two artificial intelligence models that, based on a photograph of a patient's face, estimate biological age and predict the risk of death in people with cancer.

Growtika · Unsplash License
- 0views
- 0likes and saves
- 0skips
Model trained on over 40 million images
- Researchers from Boston have created two AI models analyzing facial images: one estimates biological age, while the other predicts the risk of death in cancer patients.
- Among 122 high-risk patients, 52.5% died within 90 days, while in the lowest risk group, about 2% died.
- The authors see potential applications in assessing patients before intensive treatment, but the tool requires further research before implementation.
The Boston team built the FAHR-Face system, trained on a dataset comprising over 40 million facial photographs (Is a photo enough? This is how they want to use AI in oncology – RMF24) (AI will assess biological age based on a photo and create survival forecasts for cancer patients – rmf24.pl). Based on it, two separate tools were developed. The first, FAHR-FaceAge, assesses biological age, which is the state of the body that may differ from the number of years lived (Scientists: AI can assess the prognosis of a cancer patient based on a selfie – pulsmedycyny.pl). The second, FAHR-FaceSurvival, predicts the risk of death in cancer patients based on a single photo.
Other areas of the face
Both technologies analyze different parts of the face. The assessment of biological age is primarily based on the temples, forehead, eye area, and nasolabial fold. The algorithm predicting survival mainly focuses on the area under the eyes and the bridge of the nose (A Single Photo Could Help AI Assess How Sick a Patient Is | PetaPixel – petapixel.com). According to the authors, combining the results of both tools improved the accuracy of predictions in an external validation group of patients.
Study results
The results were published on October 19 in the journal "The Lancet Digital Health". The model identified a group of 122 cancer patients at high risk of death. Among them, 52.5% died within 90 days of assessment. Among 975 individuals classified into the lowest risk group, about 2% died during the same period.
Possible applications
The authors see this solution as a way to better determine the physiological reserve and frailty of patients before intensive treatment, such as surgery, radiotherapy, or systemic therapy. However, specialists emphasize that analyzing a photo cannot replace health assessments conducted by doctors. Further research is needed before the tool can be used in clinical practice.
Sources
- [1]AI oceni wiek biologiczny na podstawie zdjęcia i stworzy prognozę przeżywalności u chorych na raka· rmf24.pl
- [2]
Wystarczy zdjęcie twarzy? Tak chcą wykorzystać AI w onkologii· RMF24 - [3]Naukowcy: AI na podstawie selfie potrafi ocenić rokowania pacjenta z chorobą nowotworową· pulsmedycyny.pl
- [4]A Single Photo Could Help AI Assess How Sick a Patient Is | PetaPixel· petapixel.com
Discussion0
- No comments from this perspective yet. Be the first.