Imagine you work up the confidence to post a selfie you feel okay about and some freaking A.I. robot is like, “Actually, I can tell you have heart disease just by looking at this.” I mean, great, glad we detected that, but I’m also never going to take another selfie again in my life. In any case, a recent study published by the European Society of Cardiology has found that we might be able to use selfies for this purpose, potentially offering a low-cost means of getting a life-saving diagnosis just by analyzing that handsome little face of yours.
The study, published in the European Heart Journal, found that a deep-learning computer algorithm could detect coronary artery disease by analyzing four photographs of a person’s face. Per the press release, there are several facial features that tend to be associated with heart disease. Some of these are simply associated with aging, like greying hair and wrinkles, while others are a bit more challenging to immediately ascertain, like earlobe creases and small deposits of fat and cholesterol beneath the skin. However, these features aren’t regularly used to accurately diagnose heart problems by doctors.
Instead of manually analyzing these traits, researchers at the Brain and Cognition Institute in the Department of Automation at Tsinghua University in Beijing developed a computer program that could analyze a person’s blood vessels via four photographs. In a trial of the algorithm, it correctly detected heart disease in 80 percent of cases, and correctly detected the absence of heart disease in 61 percent of cases.
It’s clearly not perfect, but it’s certainly a start. While this particular study was conducted in a lab setting with photos taken by trained nurses, it’s possible that this type of technology could be used by patients at home in the future. For example, a patient could take several selfies and send them to a doctor who, using an improved version of the algorithm, could analyze the photos and determine whether the patient needs further screening. Ideally, this method would offer a low-cost and accessible way of providing care for at-risk groups.
The researchers themselves have some concerns about the ethics of such a diagnostic tool, however, and how misuse of the technology might threaten the privacy of sensitive health data. These ethical conundrums will likely only become more complicated as A.I. technology in medicine advances. In the meantime, researchers still have plenty more work to do on the algorithm, including testing it on a wider pool of patients of different backgrounds and increasing its accuracy.
For now, your selfies are just for funsies. In a few years, though, maybe they’ll save your life.