"It Could Be Useful When Medical Information Is Limited"

An artificial intelligence (AI) tool that predicts the survival chances of cancer patients from a single facial photo has been developed. It is expected to be particularly useful when medical information is limited.


Calculates Patient-Specific Risk Scores... Also Estimates Age

AI Model Predicts Cancer Patient Survival and Estimates Biological Age. Mass General Brigham Website

AI Model Predicts Cancer Patient Survival and Estimates Biological Age. Mass General Brigham Website

View original image

Bloomberg reported on October 19 (local time), citing a paper published in the medical journal The Lancet Digital Health. The AI tool was trained to estimate a person's age and predict the survival prospects of cancer patients. It was trained on more than 40 million facial images and photos of more than 30,000 cancer patients, including people who had died.


One notable feature is its ability to calculate a risk score for each patient. Patients with higher scores were found to face a greater risk of dying within 90 days.


Raymond Mak, the paper's co-corresponding author and a professor at Harvard Medical School (a radiation oncologist at Mass General Brigham Cancer Institute), said, "Doctors are trained to assess a patient's health by observing their appearance, but they are generally not very good at estimating how long a patient nearing the end of life has left to live." He explained, "I think AI can help assess a person's physiological reserve and risk of death to some extent from just a single facial photo."


The AI model estimated patients' ages with an error margin of about five years. It also classified patients into high- and low-risk groups. The two groups differed in their actual risk of death. The model was found to work effectively across patient groups with varying ages, races, genders, and cancer types.


Professor Mak said he was surprised by the AI model's strong performance in predicting survival in the study. He considered it unlikely that the tool would be used to deny or restrict patients' treatment options. Instead, noting that doctors tend to provide excessive treatment to patients nearing the end of life, he predicted that it would become "another resource for doctors to draw on."


Some Urge Caution About Clinical Use

One Facial Photo, and You're Told '90 Days to Live'... A Modern-Day Grim Reaper? View original image

The use of AI in cancer detection, diagnosis, and treatment has been growing in recent years. Its applications are also expanding, from analyzing medical imaging results to help with diagnosis to predicting cancer risk based on genetic factors and lifestyle habits.


Miroslav Bober, a professor of image processing and machine learning at the University of Surrey in the UK, who was not involved in the study, viewed the findings positively but cautioned against applying them in clinical settings without care.



"Predicting an outcome is different from explaining it," Professor Bober said, noting that "it has not been demonstrated that the tool provides clinically useful information beyond a doctor's assessment of a patient's overall health and fitness." He added that the tool would need to undergo comparative validation against doctors' own judgment, as well as validation involving other hospitals and patient groups.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing