Researchers at Mass General Brigham have developed an artificial intelligence algorithm that uses a photo of a person’s face to predict their biological age and cancer outcomes.
The tool, called FaceAge, could be effective in helping inform clinical treatment decisions in cancer care, according to a study published Thursday in medical journal The Lancet Digital Health. Researchers trained FaceAge on nearly 59,000 photos of presumed health individuals from public data sets.
Related: DaVita's Javier Rodriguez leans into AI for kidney care
While many early AI tools have focused on non-clinical use cases such as revenue cycle or documentation, there's growing interest in applying AI for diagnostics and clinical uses with oversight from doctors and nurses. Health systems such as Mass General Brigham and Mayo Clinic are increasingly testing out AI tools in clinical settings.
Along with the public data sets, FaceAge was also tested on approximately 6,000 cancer patients from two centers using photographs taken at the beginning of radiotherapy treatment. Individuals who the model predicted were older tended to be associated with worse overall survival outcomes across multiple cancer types. On average, patients with cancer had a higher biological age than those without and appeared around five years older than their chronological age, the study found.
Researchers are in the early stages of enrolling patients in a clinical trial. Hugo Aerts, director of Mass General Brigham’s Artificial Intelligence in Medicine Program and one of the study’s senior authors, said the algorithm is not ready for clinical settings.
“It nice that you have the tool, but to really help patients, you have to do a trial to really show and demonstrate that this is actually helping doctors make better decisions,” Aerts said.
The Artificial Intelligence in Medicine Program lab at MGB, which produced the study, receives National Institutes for Health grant funding. Future work has not yet been impacted recently announced cuts, a spokesperson said.
