Artificial intelligence can do a huge range of ridiculously awesome feats--parallel park a car, handle easy customer service issues or serve as your personal digital assistant, for example. But health companies are recognizing the enormous potential AI has, too. Now, Stanford University researchers might have tapped AI to deal a serious punch to skin cancer.
Developing a new system
Normally, a dermatologist starts the process of diagnosing skin cancer through a visual examination, with biopsies used for confirmation. But computer scientists from Stanford realized that this approach doesn't work in many situations, such as if a person can't afford the doctor's visit or lives far away from a clinic. So they borrowed a preexisting algorithm from Google and fed it thousands of images of skin diseases, training it to recognize which pictures were skin cancer. By the end of the project, the AI was able to identify cases of skin cancer on par with licensed dermatologists, who were asked to use traditional methods to determine whether images represented cancer.
Why this AI deserves real buzz
Currently, skin cancer ranks as the most common form of cancer, according to the Centers for Disease Control and Prevention (CDC). But as the Skin Cancer Foundation points out, the U.S. 5-year survival rate stands at 98 percent for those diagnosed early. The new algorithm could translate to more people getting properly diagnosed during the initial stages of the disease, potentially saving thousands of lives.
But the AI also has the potential to transform visual diagnostics in general. Even though tons more testing is necessary to put the algorithm into regular clinical practice, the researchers speculate that they can use the same approach to spot other conditions. What's more, they hope that, one day, they'll be able to use the AI on devices such as smartphones. These technologies are commonplace, so converting the AI would mean that access to a diagnosis becomes much less limited. And if more people can use the AI for visual diagnostics, doctors can concentrate their time on conditions and procedures where visual assessments aren't enough. That's a big deal given that, unlike in many other industries where there are a flood of qualified applicants, the health care industry is suffering for more physicians. In fact, Lenny Bernstein of The Washington Post reports that the doctor shortage might reach 90,000 by 2025. Both quantity and quality of care get addressed with the same technology.
Advances like Amazon's Alexa, high-tech security systems and movie recommendations all use AI in a practical way that makes life a little bit easier, safer or fun. But as this new research demonstrates, AI also can make life itself the target. Hats off to the computer scientists on the team, and here's to hoping their full vision comes to fruition.