AI Tool Estimates Cancer Survival Prospects from Facial Images
(MENAFN) Researchers in the United States have developed an artificial intelligence system capable of estimating cancer patients' survival prospects using a single photograph of their face, according to a study published Monday.
Created by researchers at Mass General Brigham in Massachusetts, the technology, known as FAHR-Face, examines facial characteristics to estimate biological age and the likelihood of death, potentially providing doctors with additional information when making treatment decisions.
The system was developed using more than 40 million facial photographs and subsequently adapted into two separate applications. FAHR-FaceAge estimates a person's biological age, while FAHR-FaceSurvival assesses cancer patients' survival prospects directly from facial images without using age estimates as an intermediary.
The research indicated that the two applications concentrate on different facial regions. The survival prediction model primarily analyzes areas beneath the eyes and around the nose, whereas the biological age tool focuses more heavily on features near the mouth and temples. These differences suggest that each model identifies distinct patterns potentially associated with health outcomes.
For the survival assessment, researchers trained the system using photographs from 34,363 patients before evaluating its performance on an additional 3,820 individuals. Its findings were subsequently validated independently using data from 4,892 cancer patients in the Netherlands.
The results showed that patients placed in the highest-risk category faced a mortality risk more than three times greater than those classified in the lowest-risk group. This difference remained evident even after researchers adjusted for other clinical factors.
The findings suggest that facial-image analysis could eventually provide an additional tool for assessing cancer prognosis, although further research would be needed to establish its role in clinical practice.
Created by researchers at Mass General Brigham in Massachusetts, the technology, known as FAHR-Face, examines facial characteristics to estimate biological age and the likelihood of death, potentially providing doctors with additional information when making treatment decisions.
The system was developed using more than 40 million facial photographs and subsequently adapted into two separate applications. FAHR-FaceAge estimates a person's biological age, while FAHR-FaceSurvival assesses cancer patients' survival prospects directly from facial images without using age estimates as an intermediary.
The research indicated that the two applications concentrate on different facial regions. The survival prediction model primarily analyzes areas beneath the eyes and around the nose, whereas the biological age tool focuses more heavily on features near the mouth and temples. These differences suggest that each model identifies distinct patterns potentially associated with health outcomes.
For the survival assessment, researchers trained the system using photographs from 34,363 patients before evaluating its performance on an additional 3,820 individuals. Its findings were subsequently validated independently using data from 4,892 cancer patients in the Netherlands.
The results showed that patients placed in the highest-risk category faced a mortality risk more than three times greater than those classified in the lowest-risk group. This difference remained evident even after researchers adjusted for other clinical factors.
The findings suggest that facial-image analysis could eventually provide an additional tool for assessing cancer prognosis, although further research would be needed to establish its role in clinical practice.
Legal Disclaimer:
MENAFN provides the
information “as is” without warranty of any kind. We do not accept
any responsibility or liability for the accuracy, content, images,
videos, licenses, completeness, legality, or reliability of the information
contained in this article. If you have any complaints or copyright
issues related to this article, kindly contact the provider above.
Comments
No comment