Google learned neural network to recognize skin diseases

Researchers at Google Health taught neural network to recognize pictures of skin diseases. The system simply looks at the image of the rash and makes the diagnosis. No tests are not necessary.

The neural network is able to recognize melanoma, psoriasis, acne and still more than 20 diseases, up to 93%. The algorithm takes into account not only the image, but also to 45 other indicators, including medical history, symptoms, age and gender. The system handles all the indicators, compares them with its database and gives the three most probable diagnosis.

Scientists trained neural network is 17.7 thousand pictures of diagnosed skin diseases. they also used more than 50,000 diagnoses that have delivered more than 40 dermatologists.

Google has yet to bring the system up to the mind. But one day she will be able to quickly identify and sort almost any skin diseases. And the doctors will have more time on patient care.