As we creep ever close to the Singularity, the moment when an artificial intelligence becomes self-aware, machine learning is still a mostly benign and helpful force in our lives. Until the Terminator feet start stomping skulls in the post-nuke wasteland, we have a lot to be gained from our computer friends.
A new machine learning tech developed by Google and Verily can allegedly predict cardiovascular health risk in patients based only on a photograph of their eye. The test is also allegedly as accurate a predictor of major cardiovascular incident as traditional medical tests. With just a picture of your eyeball, the program can tell whether you smoke, what your blood pressure is and how likely you are to have a heart attack.
Not just any photograph will work. The program needs a photo of the blood vessels inside the eye's lining, called the "retinal fundus." It compares the photo against a database of other fundus pics that are associated with medical profiles in order to make a prediction.
YourStory via Flickr
The medical community is excited about the new invention. As it stands, it stands toe to toe with more traditional (read: complicated and expensive) predictive blood vessel scanning techniques. It is likely to eventually surpass those methods in accuracy and speed, as well.