Can AI predict whether you’ll die from cancer from just an image of your face?
Researchers tasked an algorithm with predicting cancer patients’ risk of dying. For some people, it was shockingly accurate
Can AI accurately predict a cancer patient's risk of dying simply by analyzing a photograph of their face? Researchers from Mass General Brigham Cancer Institute have developed an artificial intelligence model that can make this prediction, according to new research published in the journal Lancet Digital Health. While not perfect, the AI model found that patients with higher risk-of-death scores tended to have shorter lifespans.
The algorithm builds on the way human doctors assess their patients' health, says Ray Mak, a radiation oncologist who led the research. "We're trained in medical school to walk into a room, look at the patient and make an assessment—an impression of how healthy, frail or sick they are," Mak explains. However, this assessment is subjective.
The AI, trained on 40 million images of faces, was then tweaked to analyze new photographs in two ways: to estimate a person's biological age and to predict their odds of survival. The model proved surprisingly accurate, particularly when predicting death risk in cancer patients. In a test group of 3,820 cancer patients, 64 out of 122 patients who received the highest risk scores died within 90 days of their photograph being taken.
A separate group of over 38,000 cancer patients also showed a correlation between looking older than one's chronological age and a higher risk of dying. The AI tends to focus on specific facial areas when making these predictions, including the nasolabial folds, the forehead, and areas below the eyes. So far, the AI has shown promise, but it is not yet ready for clinical use.
The researchers are working to expand their training sets to include a more diverse range of faces, currently the AI has only been trained on a Boston-area cancer patient population that was predominantly white. Ethical considerations also arise with this technology, as it raises concerns about insurance companies and employers potentially using it to deny coverage or discriminate against individuals based on their perceived health. Despite these concerns, Mak remains optimistic about the potential of this new field in medicine.
Written by urgent.news from Scientific American's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
