An AI model trained on more than 40 million facial images is moving the discussion about preoperative assessment into unfamiliar territory.

On October 19, Mass General Brigham reported research on FAHR-Face, a foundation model used to develop two tools. FAHR-FaceAge estimates biological age from a photograph. FAHR-FaceSurvival predicts survival risk in patients with cancer. The study appeared in The Lancet Digital Health.

The researchers found that the difference between estimated biological age and chronological age was independently associated with survival. The two tools captured partly different information, and combining them improved survival-model fit in an external validation cohort. The 40 million images describe the foundation model’s training scale; they do not represent 40 million surgical patients.

The potential relevance to surgery is clear. Two patients with the same birth date can have very different physical reserves. Another measure could eventually help clinicians discuss treatment intensity and recovery expectations.

That possibility is also where the editorial caution belongs. This research does not establish that a photograph can determine whether someone is suitable for an ASC, replace an anesthesia evaluation or predict complications across orthopedic, spine and cosmetic procedures.

Before buying such a product, a surgical organization should ask which patient population was used for validation, what outcome the tool predicts and whether it improves decisions beyond information already available. A statistically significant association is insufficient justification for changing a patient’s treatment.

There are practical questions, too. Who obtains consent for the image? Where is it stored? Can a patient decline? How would a clinician explain a risk estimate that conflicts with the examination?
For now, this is research to follow rather than a scheduling criterion. Its most interesting contribution is the suggestion that ordinary photographs contain measurable health information. Whether that information improves surgical care requires studies designed around surgical decisions and patient outcomes, with clear evidence that clinicians can act on the results.